# Kula Blog: full content > Complete text of published Kula blog posts. Kula makes software for boutique fitness studios; Kula Intelligence is its read-only AI layer, delivered as an MCP server at https://mcp.kula.digital/mcp (see https://kula.digital/kula-intelligence/mcp). Docs: https://docs.kula.digital # The four parts of a Gen 2 fitness business: Intelligence, Tribe, apps and AI workers URL: https://kula.digital/blog/gen-2-four-parts Published: 2026-09-09 TL;DR: A Gen 2 fitness business puts one shared, governed understanding of the business underneath everything else, read from the systems it already runs. Four parts sit on top: Kula Intelligence (ask the business through Claude, ChatGPT or Gemini, available today for Mindbody, Wix and GymMaster), Tribe (the instructors' app, available with Kula Fitness), apps built on the shared context (examples and reference builds), and AI workers acting inside policy and permissions (the direction). Access at every layer depends on supported connections, policy and permissions. Every fitness business runs on a stack of systems that were never designed to talk to each other. Bookings in one. Payments in another. Payroll, accounting, marketing, support, each with its own login and its own idea of who your members are. The person joining it all together is the owner, and the joining happens in their head. A Gen 2 business puts that shared understanding into software, once, underneath everything else. This post is the map: the four parts that sit on top of it, what each one is for, and an honest line between what you can use today and where it is going. ## The one idea underneath Kula Intelligence connects across the systems you already run and builds one governed model of the business from them: members, bookings, attendance, staff, classes, payments, documents. It does not replace any of those systems and it does not ask you to migrate. It reads them, through safe views rather than raw access, and keeps the model current as the day goes. That model is the shared context. Everything below draws on it, subject to policy and permissions. That last clause matters. What an instructor can see is not what an owner can see, and neither is what an AI worker would be allowed to act on. ## Part one: Intelligence. Ask it. The first thing you do with a shared understanding is ask it questions. Kula Intelligence is delivered as an MCP server, the plumbing that lets Claude, ChatGPT or Gemini reason over your business the way they reason over a document you paste in. You ask in plain language, in the app already on your phone, from wherever you happen to be standing. The teacher who resigns on a Tuesday is the example we use because it crosses systems: who could take the classes, which members are anchored to that teacher, what to do first. One question, several systems, no prebuilt workflow. You can try it on a real studio's data at [kula.digital/ask](/ask). Available today for studios on Mindbody, Wix or GymMaster. Other booking systems are on the connector roadmap, and telling us yours moves it up the list. ## Part two: Tribe. Put the context with the team. The owner is not the only person who needs context. Instructors need it more, and they need it at 5:55am, not in a report. Tribe is the teachers' app. Schedule across every studio they teach at, one login. Whispers before class: "Sarah's back after three weeks, welcome her home", with the reason attached, and the teacher still deciding what to say. Messages in one place. Covers offered to the people who are available and qualified, first to accept gets it, and the booking system updated. Compliance documents tracked. Tribe is where the shared understanding reaches the floor. It is a prompt for human judgement, not a script. Tribe ships with Kula Fitness and is available today. Details at [the Tribe app](/kula-fitness#tribe). ## Part three: Apps. Build what you need. Every studio has three or four things it does its own way. Instructor bonuses that reward the classes that fill. PT commissions. An intro-pass tracker that tells you who is about to lapse before the pass runs out. Off-the-shelf software either does not have them or has a version that does not fit. Because the business model already exists underneath, those tools can be built on top of it rather than starting from raw exports. Our [Instructor Bonus Builder](https://github.com/kulaforbusiness/instructor-bonus-builder) is published on GitHub with full documentation and a live demo, so you can run it yourself or build your own from it, and the instructor bonus, PT commission and intro-pass examples we show at AusFit are exactly that: examples built on the business context, by people who know the business. This part is the direction, not a catalogue. What you can build depends on your systems, your data and what you want the tool to do, so tell us the job and we will tell you what is possible now and what needs work. ## Part four: AI workers. Give AI work. The last part follows from the first three. Once the business is understood in one place, some of the work that draws on that understanding can be handed to an AI worker: the weekly read of what needs attention, the draft of the message to a member who has gone quiet, the first pass at next month's roster. Humans and AI working from the same understanding of the business, with the human making the call. We are deliberate about this one. An AI worker acts inside the same policy and permission model everything else uses, every action is on the record, and nothing is sent or changed without approval where it matters. This is the direction of Gen 2 and we will show it as that: where it is going and what we are building towards, not something on the price list today. ## The stack, in one picture | Layer | What it is | Status | |---|---|---| | Your AI, in your hand | Claude, ChatGPT or Gemini, asking the business | Available | | Tribe | The instructors' app: schedule, Whispers, messages, covers, compliance | Available with Kula Fitness | | Apps | Tools shaped by your business, built on the shared context | Examples and reference builds; talk to us | | AI workers | Work handed to AI, inside policy and permissions | Direction | | Kula Intelligence | One governed model of the business, read from the systems you already run | Available today for Mindbody, Wix and GymMaster | | Your systems | Bookings, payments, payroll, accounting, marketing, support | Unchanged | Access at every layer depends on supported connections, policy and permissions. ## What to do with this Start at the top of the stack. Bring one question you cannot answer without three logins, tell us which booking system you run, and see what a connected business looks like. [Ask your whole business](/ask), or find us on Stand 65 at AusFit 2026. --- # One level at a time: how a studio actually becomes Gen 2 URL: https://kula.digital/blog/gen-2-one-level-at-a-time Published: 2026-09-09 TL;DR: A studio becomes Gen 2 in steps, not a migration: connect the business you already have (about thirty minutes), ask it the same day, put intelligence with the team through Tribe when the team is ready, build a tool when one is needed, and hand work to an AI worker last. Each step uses the context the last one built, nothing below is replaced, and you can stop at any level. The model you ask in commoditises; the understanding of your business compounds. Every conversation about AI in a studio eventually hits the same wall: it sounds like a project. A migration. A month of someone's life and a risk to the thing that pays the bills. I want to take that wall down, because the way a studio becomes Gen 2 is the opposite of a project. It is a sequence of small steps, each of which pays its way before you take the next. ## Start with the business you already have A working studio: people above, the software they already use below. Owner, manager, teachers, PTs, members. Bookings, payments, CRM, accounting, marketing, payroll. Calm, not broken. That is the starting point, and it persists to the end. Nothing below gets replaced at any step. ## Step one: connect it Pick your booking system and connect in minutes. The Kula layer slides in between the people and the software. A single studio is usually ready in about thirty minutes, and we email you the moment it is. Bigger multi-site books take a few hours. Mindbody, Wix and GymMaster today; tell us yours if it is not on the list. ## Step two: ask it The same day, ask the questions that used to take three logins, in Claude, ChatGPT or Gemini, from wherever you are. This is where most studios start, and where many stay for a while. That is not a failure to progress. It is the level doing its job. Underneath, the shared memory is compounding: every answer, every correction and every outcome becomes context. ## Step three: put intelligence with the team When the team is ready, add Kula Fitness and give your instructors Tribe. Same context, now on the floor, before class. There is nothing to migrate because the context your instructors get is the one you have already been asking questions of. ## Step four: build what you need When a tool you wish existed comes up, describe it. Build one on the business, use it, then decide on the next. You will know you are here when you catch yourself saying "I wish something just told me when…" about the third thing that month. ## Step five: give AI work When there is a job you would hand to a capable new hire with clear rules, hand it to an AI worker with the same rules, inside the same policy and permissions, with approval where it matters. One job at a time. This is the direction, and it comes last because it needs everything before it. ## You can stop at any level This is the sentence I want every owner to hear. Each level uses the context the one before it built. Stop after step two and you are running a Gen 2 studio at level one, and it is already a different business than the one where you were the integration, the memory and the decision engine. Nobody is behind for staying at a level. Nobody has to do all four. ## Why the memory matters more than the model Models commoditise. Business understanding compounds. The AI you ask questions in will get better every few months whatever you do. The thing that gets more valuable with time is the understanding of your business that Kula holds: what actually happened, why a choice was made, what worked next time. It stays when people leave. Whichever level you are on, that memory is working for the next one. ## How to begin Bring one question you cannot answer without three logins, and tell us which booking system you run. Put the question to [The Ask](/ask) on a real studio's data first if you like. Then connect yours. Level one is thirty minutes away, and it is the only level you have to take today. --- # Level four of a Gen 2 studio: give AI work URL: https://kula.digital/blog/gen-2-level-four-give-ai-work Published: 2026-09-09 TL;DR: Level four of a Gen 2 studio hands work to AI workers: the weekly read of what needs attention, drafts for quiet members, a first pass at the roster. An AI worker is only as good as the business it can see, so these workers stand on the same shared understanding as the owner, the instructors and the custom apps, inside a five-layer policy engine, with every action on the record and approval where it matters. It is the direction Gen 2 is building towards, not a catalogue on the price list. The first three levels of a Gen 2 studio are about people: the owner asking, the instructors knowing, the operator building. Level four is about the work itself, and who, or what, does it. ## From answering questions to doing work AI is moving along a line. A chat window answers. A model that reasons understands. A model that acts, acts. A workforce owns jobs. The step-change is not a better chatbot. It is software that can reason across the business and complete work. For a studio that means the weekly read of what needs attention, done before you ask. The draft of the message to a member who has gone quiet, waiting for a human to send. The first pass at next month's roster. A retention worker, a marketing worker, a finance worker, an operations worker, a member support worker, a compliance worker, taking their place in the same row as the owner, the manager, the teachers and the PTs. The Gen 2 workforce is human plus AI. ## The gap most AI workers fall into An AI worker is only as good as the business it can see. Point a marketing agent at your ad account and it sees ads. Point a finance agent at your books and it sees invoices. Point a retention agent at your CRM and it sees the CRM's idea of a member, which is not the same as the booking system's idea of that member, which is not the same as the payment platform's. Each one sees a slice, and each one can be confidently wrong about the rest. Another silo does not fix the silos. That is why level four sits on the same shared understanding as the other three. One business, one shared context, many AI workers. Everyone, human and AI, works from the same business understanding, subject to the same policy and permissions. ## Policy, permissions and the record This is the part I care most about getting right, because handing work to software is different from asking it questions. An AI worker in a Gen 2 studio acts inside the same five-layer policy engine everything else uses: tenant isolation, role scope, relationship scope, consent scope and purpose limitation. It sees what its job allows and nothing else. Every action is on the record. And where an action matters, nothing is sent or changed without a person approving it. The difference from a human worker is not that the AI needs fewer rules. It is that the rules are enforced by the layer it stands on, not by a policy document it might forget to read. ## Where this level actually is Level four is the direction of Gen 2. We show it as that: what we are building towards, with the shared understanding, the policy engine and the record already in place underneath, and not as a catalogue of workers on the price list today. If you have a job you would give a capable new hire with clear rules, we want to hear about it, because those jobs are the ones that shape what gets built first. ## Where it sits on the path You get here last, and only when the earlier levels have done their work. A studio that has been asking questions for months, has its instructors on Tribe and has built a tool or two has something an AI worker can stand on: a business that is understood, with memory of what happened and what worked. That is the whole reason the levels are in this order. The series closes with [one level at a time](/blog/gen-2-one-level-at-a-time), on how a studio actually gets here. --- # Level three of a Gen 2 studio: build what you need URL: https://kula.digital/blog/gen-2-level-three-build-what-you-need Published: 2026-09-09 TL;DR: Level three of a Gen 2 studio is operators building the tools their business needs on the shared understanding that already exists underneath: an instructor bonus model, PT commissions, an intro-pass tracker and more. AI writes the code; studio experience decides what should exist. These are examples built on business context, not included modules, and each one starts with a conversation about the job. I have run studios for twenty-five years. I am not a software engineer. This year I started building the software my studio needs, and I want to be precise about why that is possible now and what it does and does not mean. ## The tools every studio wishes existed Every studio has three or four things it does its own way, and off-the-shelf software either does not have them or has a version that does not fit. Mine were an instructor bonus model that rewards the classes that actually fill, measured against the true capacity of the time slot rather than the size of the room. PT commissions. A small-class notice so a teacher knows before they walk in. An intro-pass tracker that flags who is about to lapse before the pass runs out. A Meta tracker. An invoice creator. In a Gen 1 studio the path to any of those was: problem, request, specification, roadmap, developer, integration. Six steps, most of them someone else's, and a year later you have a feature that almost does what you asked. The Gen 2 path is: problem, describe, build, use. ## AI writes the code. Experience decides what should exist. The code is not the hard part any more. Claude Code, Codex, Gemini and the rest will write it. The hard part is knowing what should exist, and twenty-five years of standing in studios is the only thing that answers that. A software engineer would not know that a 6am slot that realistically holds twelve should be judged against twelve, not against a room built for twenty-five. I do, because I have watched compensation punish good teachers for the timetable for two decades. ## Build on the business, not another silo Here is the part that makes it Gen 2 rather than a hobby. Every one of those tools is built on the business context that already exists underneath: members, classes, teachers, payments, operations, read from the systems the studio already runs. Nobody exports a CSV. Nobody re-enters anything. The instructor bonus model reads the same attendance the owner has been asking questions of at level one and the same schedule the instructors see in Tribe at level two. That is what "build what you need" means. Not that AI can build apps, which everyone knows by now, but that the people closest to the problem can build on a shared understanding of the business instead of beside it. Software at the speed of the operator. If you want to see exactly what one of these looks like, the [Instructor Bonus Builder](https://github.com/kulaforbusiness/instructor-bonus-builder) is published on GitHub with full documentation, including a fully worked month, and a [live demo](https://instructor-bonus-builder.vercel.app) that needs no login. You can run it on your own studio or gym, or build your own from it. A deterministic engine splits a fixed pool on retention and capacity, reconciled to the cent, and every allocation opens a panel that explains the number line by line. The AI explains the maths. It never decides who gets paid what. ## What this level is, honestly These are examples built on the business context, not included modules, and not a catalogue. What you can build depends on your systems, your data and the job you want done. So level three starts with a conversation: tell us the job, and we will tell you what is possible now, what needs configuration and what still needs work. Where it sits on the path: you get here when a tool you wish existed comes up, not before. Build one, use it, then decide on the next. And once the business is understood well enough to build on, it is understood well enough to hand work to, which is [level four](/blog/gen-2-level-four-give-ai-work). --- # Level two of a Gen 2 studio: put intelligence with the team URL: https://kula.digital/blog/gen-2-level-two-intelligence-with-the-team Published: 2026-09-09 TL;DR: Level two of a Gen 2 studio hands the shared understanding of the business to instructors through Tribe: their schedule across every studio, who is coming, Whispers with the reason attached, messages, covers that update the booking system, and compliance. Whispers are prompts for human judgement, not scripts. Tribe ships with Kula Fitness and is added when the team is ready. Level one gives the owner a way to ask the whole business. Level two hands the same understanding to the people who actually stand in the room. ## The problem level two solves Your instructors walk into a class of thirty people knowing who is booked and not much else. The context that would change how they teach that hour lives somewhere else: who is back after three weeks away, who is on their last intro-pass session, who has quietly gone from four visits a week to one. In a Gen 1 studio that context reaches the floor in a report, later, if someone runs it. The business brain is only useful when it reaches the person at the moment it matters. That moment is 5:55am, on a phone, not in a dashboard. ## What Tribe is Tribe is the instructors' app. It brings the teaching day into one place: where and when they are teaching, who is coming before they arrive, Whispers, messages, covers and compliance, across every studio they work at. Whispers are the part people remember. A Whisper is a quiet nudge that lands with the instructor before class, with the reason attached. "Sarah's back after three weeks, welcome her home." "New face, three classes in, learn her name." It is a prompt for human judgement, not a script. The instructor still decides what to say, and the connection stays between people. Covers work the way covers should. When a class needs covering it is offered to the people who are available, qualified and a good fit. The first to accept gets it, the booking system updates itself, and both the instructor and the manager are confirmed. If nobody picks it up, it escalates, so a class is never left uncovered. And for the instructor who teaches at three studios, one login. Every class in one schedule, covers from any studio in one place, messages and documents together instead of a different app and group chat per studio. That is a large part of why teachers value it. ## Less time operating software. More time with people. That is the whole point of level two. Nothing about it adds admin to an instructor's day. It removes the part of the day spent hunting for context and hands the context over, so the hour in the room is spent on the people in the room. ## Where it sits on the path Tribe ships with Kula Fitness, the studio side owners and managers run the business from. You add it when the team is ready, not before. Because it works from the same shared understanding as level one, there is nothing to migrate and nothing to re-enter: the context your instructors get is the context you have already been asking questions of. Read more at [the Tribe app](/kula-fitness#tribe). When a tool you wish existed comes up, that is [level three](/blog/gen-2-level-three-build-what-you-need). --- # Level one of a Gen 2 studio: ask your whole business URL: https://kula.digital/blog/gen-2-level-one-ask-it Published: 2026-09-09 TL;DR: Level one of a Gen 2 studio is Kula Intelligence: connect the systems you already run, keep them exactly as they are, and ask the whole business in the AI you already use, from anywhere. A single studio is usually answering questions in about thirty minutes. Most studios stay at this level for a while, and the shared memory compounds the whole time. Every Gen 2 studio starts the same way, because the first level is the one that does the most for the least effort. You connect the software you already run, and you ask the business what you want to know. ## What it is Kula Intelligence connects across the systems you already run, builds one governed model of the business from them, and gives the AI you already use, Claude, ChatGPT or Gemini, that model to reason over. You ask in plain language, in the app already on your phone, from wherever you happen to be standing. On the floor. Between classes. In the car. Nothing changes in the systems your team uses every day. Kula reads them through safe views rather than raw access, and it writes nothing back. Your booking system is still your booking system. Your accountant still has Xero. You just stop being the only thing that joins them together. ## The example we keep coming back to A senior teacher resigns on a Tuesday. In a Gen 1 studio that is an evening of logins: who could take her classes, which members only come for her, what the roster looks like next week, what it costs. In a live session with a studio manager, the same question went to Kula with the business context already connected. Back came replacement options, the members anchored to that teacher and what to do first, from one question, with no prebuilt workflow. We told the whole story in [episode one](/blog/gen-2-business-ep-01-teacher-resigned). You can put a version of that question to a real studio right now. [The Ask](/ask) is a live conversation with Juniper Studio, a real studio with its name changed and everything else left as it is. ## What to ask first The questions that work best at level one are the ones that used to take three logins. A few we hear at every stand and every onboarding: - How are we going today? - Who needs my attention this week? - What happened while I was away? - Which teacher keeps new members, and which quietly loses them? - Where am I turning people away? Ask them the way you would say them out loud. No syntax, no filters, no dashboard to learn. ## How long it takes Connecting takes a few minutes. A single studio is usually answering questions in about thirty minutes, and we email you the moment it is. Bigger multi-site books can take a few hours. Mindbody, Wix and GymMaster are connected today. If you run something else, tell us, and it moves up the connector roadmap. ## Why most studios stay here for a while Level one is not a stepping stone you rush past. Most studios stay here for weeks or months, asking better questions each week, and that is exactly right. Two things are happening underneath. You are learning what your business actually looks like when the silos are gone. And the shared memory is compounding: every answer, every correction and every outcome is context the next level will use. When the team is ready for the same context on the floor, that is [level two](/blog/gen-2-level-two-intelligence-with-the-team). --- # Today, the owner is the business brain. Gen 2 is what changes that. URL: https://kula.digital/blog/gen-2-owner-is-the-business-brain Published: 2026-09-09 TL;DR: In a Gen 1 studio the owner is the business brain, joining bookings, payments, payroll, accounting and marketing together by hand. AI is moving from answering questions to doing work, but an AI worker is only as good as the business it can see, and another silo does not fix the silos. A Gen 2 business puts one shared understanding of the business underneath the AI, the team, custom apps and AI workers, read from the systems the studio already runs. It is a path taken one level at a time, not a migration. Ask any studio owner a real question about their business and watch what happens. How did last week go? Which teacher keeps members? Can we afford another 6am? The answer exists. It is just spread across the booking system, the payments platform, the payroll run, the accounting package, the marketing tool and a spreadsheet somebody built two years ago. So the owner says the sentence every owner says: let me check and get back to you. That sentence is the whole problem. Bookings, payments, payroll, accounting, marketing and spreadsheets all hold pieces of the truth. None of them holds the whole. So the owner becomes the integration, the memory and the decision engine for the business, usually at night, usually alone. > We call the software that got us here Gen 1. It recorded the business. It never understood it. ## Why now AI is moving from answering questions to doing work. A chat window gives you answers. A model that can reason understands. A model that can act, acts. A workforce owns jobs. The step-change is not a better chatbot. It is software that can reason across the business and complete work. But there is a catch that every studio will run into. An AI worker is only as good as the business it can see. Point a marketing AI at your ad account, a finance AI at your books and a retention AI at your CRM, and each one sees a slice. Each one can be confidently wrong about the rest. Another silo does not fix the silos. ## Why the alternatives miss the same layer The AI built into your booking system sees one system. A new all-in-one platform asks you to migrate, which means moving years of process, training and data so that one vendor can see everything. A consultant builds one client at a time and does not scale. All three miss the same thing: the layer underneath the tools, where the business is understood once and shared with everything that needs it. ## What a Gen 2 business is A Gen 2 business puts one shared understanding of the business underneath everything else. It is read from the systems you already run, so nothing is replaced and nothing is migrated. Your AI works from it. Your team works from it, on their phones, before class. The tools you build work from it. The AI workers you eventually hire work from it, inside the same policy and permissions as everyone else. Four things sit on top of that understanding, and they are the four levels of Gen 2: ask it, put intelligence with the team, build what you need, give AI work. We wrote the map of all four in [the four parts of a Gen 2 fitness business](/blog/gen-2-four-parts). ## It is a path, not a migration This is the part I want to say loudest, because I have run studios for twenty-five years and I know what "transformation" sounds like from the other side of the desk. Nobody moves a studio to Gen 2 in a weekend, and nobody has to. You start with the business you already have. You connect it, which takes about thirty minutes for a single studio. You ask it, the same day. Most studios stay there for a while, and that is fine, because the shared memory is already compounding. When the team is ready, you put intelligence with the team. When a tool you wish existed comes up, you build it. When there is a job you would give a capable new hire with clear rules, you give it to an AI worker with the same rules. Each level uses the context the last one built. You can stop at any level and still be running a Gen 2 studio at that level. The next post in this series takes the first step: [level one, ask your whole business](/blog/gen-2-level-one-ask-it). --- # Ask your whole business: Kula Gen 2 at AusFit 2026 URL: https://kula.digital/blog/kula-gen-2-ausfit-2026 Published: 2026-09-08 TL;DR: At AusFit 2026 Kula is introducing the Gen 2 fitness business. Kula Intelligence gives the AI an operator already uses (Claude, ChatGPT or Gemini) a shared understanding of the whole business across the systems they already run, so any question can be asked in plain language from anywhere. Tribe brings the same context to instructors: schedule, Whispers, messages, covers and compliance in one app. The live demo is at kula.digital/ask and the team is on Stand 65 at ICC Sydney on 11 and 12 September. If you run a studio or a gym, you know the question that sends you back to the computer. How did last week go? Who has not been in for a while? Can we afford another teacher on Thursday mornings? The answer exists. It is just spread across your booking system, your payments, your accounting and a spreadsheet you built two years ago. You are the one who joins it together, usually at night. This week at AusFit 2026 we are introducing what we call the Gen 2 fitness business. The idea fits on a flyer, and it did: your whole business, one conversation away. ## Keep your software. Use your AI. Ask your whole business. Gen 1 software recorded your business. Each system knows its own part and nothing else. Your booking system knows bookings. Your accounting package knows invoices. Your marketing tool knows who opened an email. None of them knows what a teacher leaving means for the members who only come for her. A Gen 2 business changes one thing. It gives the AI you already use, Claude, ChatGPT or Gemini, a shared understanding of the whole business, across the systems you already run. You keep the software. You keep the AI. Kula Intelligence connects the two, so you can ask a question in plain language and get an answer that draws on more than one system. From your phone, on the floor, between classes. We wrote about the thinking behind this in [The Gen 2 studio](/blog/gen-2-studio-help-on-every-decision). AusFit is where it becomes something you can try. ## A senior teacher resigns on a Tuesday The example we keep coming back to happened in a live session with a studio manager, mid-conversation, when a teacher texted to resign. We told that story in [episode one](/blog/gen-2-business-ep-01-teacher-resigned). The short version: with the business context already connected, one question turned into replacement options, the members most at risk from the change and the next actions. No prebuilt workflow, and nobody opened a spreadsheet. You can ask a version of that question yourself today. [The Ask](/ask) is a live conversation with Juniper Studio, a real studio with its name changed and everything else left as it is. Tap "If my top teacher left?" or "How are we going today?" and watch the answer come from the studio's own data. Then ask it something of your own. ## Tribe: the same context, in your instructors' hands Owners are not the only people who have to hold the business in their heads. Your instructors walk into a room of thirty people knowing who is booked and not much else. Tribe is the app your teachers and instructors open. Their schedule across every studio they teach at. Who is coming before they arrive. Whispers: a quiet nudge about a member who may need a connection, and why. Messages in one place instead of four group chats. Covers they can pick up in a tap. Their compliance documents, current. The point is not more software for teachers. It is less time operating software and more time with people. Tribe ships with Kula Fitness. Read more at [the Tribe app](/kula-fitness#tribe). ## Where this goes next The shared understanding under the AI and under Tribe is the part we think matters most, because it is what everything else builds on. Tools shaped by the people who know the business. Work handed to AI that draws on the same context. That is the direction of Gen 2, and we will show it at the stand as exactly that: where it is going, not something to buy today. The four parts, and what you can use now, are in [the next post](/blog/gen-2-four-parts). ## Come and find us We are on Stand 65 at AusFit 2026, ICC Sydney, on Friday 11 and Saturday 12 September. Bring a question about your own business, the one that takes three logins to answer, and tell us which booking system you run. We will put it to the demo with you and be straight about what Kula can do for your setup today. If you cannot make it, the demo does not need a stand. Ask your whole business at [kula.digital/ask](/ask). --- # You are the glue URL: https://kula.digital/blog/you-are-the-glue Published: 2026-08-11 TL;DR: Studio software splits the business across five or six systems that never share context, and quietly assigns the operator the job of joining them at a desk. The real cost is the floor conversations that never happen: the drifting member nobody catches, the teacher whose question waits for a schedule check. Kula Intelligence builds one model of the business underneath the systems you already run, so one spoken question returns one joined answer wherever you are standing, and the conversation happens while the person is still in the room. It is 9:40 on a Tuesday and you are exactly where you should be: on the floor. A member stops you on her way out. She wants to pause her membership while she travels, and she wants to know what happens to the two weeks she has already paid for. The answer exists. It lives in the office, behind a login, split across two systems that have never spoken to each other. So you say the sentence every operator says a dozen times a week. "Let me check and get back to you." ## You are the glue A studio runs on five or six systems. Bookings in one. Payments in another. Payroll, accounting, marketing, messages. Each is competent at its own job. None of them share what they know. Any question that spans two of them gets answered by a person walking to a desk and lining up tabs. The industry never built the integration layer. It quietly assigned the job to you. And the desk is expensive in a way the clock does not show. This business is built standing up: the welcome back after three weeks away, the chat after class that turns a casual into a regular, the ninety seconds before a class starts. Members rarely leave loudly. They drift, and drift gets caught by whoever is on the floor talking. Every trip to the office is a conversation that never happened, and some of those conversations were worth a membership. ## The other version of the day Here is the same morning with the glue removed. It is 9:58 and you are walking to studio two. You say, out loud, to the earbud you are already wearing: "Hey Kula, I am about to take Maya's class. What should I know?" By the time you reach the door, this comes back. Spoken, not shown. > Twelve in the class. Debbie hits her 200th class today. > > John is back after three weeks away, but he is drifting. On his current pattern he is likely gone in two months. His remaining lifetime value is about $450. > > Maya carries around $100K a year in revenue at a 25% margin. She ranks sixth on performance and second on member connection. She has asked twice about taking on more classes. Talk to her after class. One question. Four systems: bookings, payments, accounting, communications. Read, joined, ranked and spoken in the time it took to walk there. No login. No tabs. No trip to the office. ## What you do with ninety seconds You open the class by celebrating Debbie's 200th in front of the room. It costs nothing and lands forever. You catch John on the way out. Not with a win-back email three weeks after he cancels, but with "good to see you back, we missed you," and a real conversation about what got in the way. That conversation is how drift ends. On the numbers above, it is the difference between $450 of lifetime value and roughly double that. ## And then Maya Maya is the one most systems would never flag at all, because she is not a risk. She is an asset. She wants more classes, and the usual answer is the desk answer: "Let me look at the schedule and get back to you." Instead, you stay where you are standing and dig deeper, with her. "Which slots could Maya realistically take?" Tuesday 6am and Thursday 7pm are running under capacity and have been patched with covers for six weeks. "What happens if she takes them?" Her regulars follow her. She is second on connection, and when her classes move, her members move with them. Two more classes puts her among the top teachers by revenue carried. Now the conversation about her future happens while she is still in the room, with real numbers underneath it. She asked for more classes. What she gets is a path: which classes, why those, and what it means for her income and her standing. Teachers leave when they cannot see a future. This one just saw hers, in the two minutes after class, because you did not have to say "let me get back to you." ## The glue becomes software Everyone else in this industry is bolting an AI assistant into their own box. The booking platform gets one, the payments platform gets one, each smart about its own data and blind to the rest. That leaves you exactly where you were: the glue, now between smarter boxes. Kula Intelligence takes the other route. It builds one model of the business underneath the systems you already run: who booked, who paid, who taught, who was messaged, and how it all connects. Then it answers across all of it, wherever you are standing, and lets you keep asking. Nothing to replace on day one. In a business built on people, the measure of software is not how good its screens are. It is how rarely you need them. Your time belongs to the room, and to the Debbies, the Johns and the Mayas in it. If you want to feel it, ask a question at [kula.digital/ask](https://kula.digital/ask). --- # The filing cabinet finally does the work URL: https://kula.digital/blog/the-filing-cabinet-finally-does-the-work Published: 2026-08-03 TL;DR: Lassie, an a16z-backed AI company automating dental back offices, built its context layer and tools before models were good so when reasoning models arrived, the product got smarter overnight. That's the substrate thesis proven in another industry: the model is rented, the substrate compounds. Kula Intelligence applies the same architecture to fitness, with one different bet instead of building a closed agent, Kula builds the governed data layer and lets studio owners use the assistant they already have (Claude, ChatGPT, Gemini) through the ask model, and puts the same substrate in teachers' hands through Kula Tribe. a16z recently published a conversation with Steijn Pelle and Frédéric Renken, the founders of Lassie, an AI company that runs the financial back office for more than 700 US dental practices. Alex Rampell led their $35M Series A and led the discussion. It's the best hour on AI for small business I've heard this year, and nearly every minute of it maps onto what we're building at Kula for studios and gyms. Here's what they said, where we agree, and where we've made a deliberately different bet. ## Software stored the work. People still did it. Rampell opens with a history lesson. The origin of software was taking filing cabinets and putting them in databases. Airline reservations became Sabre. HR filing cabinets became PeopleSoft. Accounting cabinets became QuickBooks. Storage went digital. The work stayed human. An HR department in 2000 wasn't meaningfully smaller than one in 1950, the files just moved from a locked drawer to a locked server. A dentist could see every overdue invoice in the system. Chasing them was still a person's job. Steijn's own dentist, the top-rated practice in his area, was spending 200 hours a month on paperwork. The claim at the heart of the episode: AI is the first generation of software that does the work instead of storing it. Lassie's numbers back the claim. Their first agent hands a practice back around 30 hours of labor a month, autonomously, with payment posting at 98% automation. If you run a studio or a gym, this arc is yours too. Your booking platform, your payment processor, your access control system, your website — they digitised the records of your business. Understanding what's actually happening in it — who's about to leave, which classes make money, whether the intro offer converts — stayed manual. Usually it's the owner. Usually at night. ## The part of the interview that mattered most Lassie started in 2020, before the models were good. So they built the two things an agent needs regardless of the model: context, meaning access to all the historical data of the practice, and tools, meaning the integrations to actually act. The intelligence layer wasn't intelligent yet — early on, the founders were literally the humans in the loop, processing payments by hand. Then the models got good. And because the context and tools were already built, they could swap better intelligence straight in. The product got smarter almost overnight, on someone else's R&D budget. That is the substrate thesis in a single anecdote. The model is not the moat and it is not the product. Models are rented. They improve on someone else's schedule and they're available to everyone, including your competitors. The durable asset is the layer that makes a model useful: connected systems, one governed definition of the business, and safe tools to act. The substrate wins. ## Models don't know your business The second insight is one we've written about before, so it was striking to hear it from a company in a completely different industry. Frontier models are trained on effectively the whole internet and they still can't do this work, because the workflows aren't on the internet. They live in the heads of office managers and in actual filing cabinets. Frédéric assumed the reasoning models would just know how to bill an insurance claim. They don't. Worse, every system in the ecosystem defines things slightly differently. Lassie had to build an ontology so that every system agreed on what an insurance claim and a patient payment actually are before an agent could safely touch anything. Fitness has the identical problem. A "member" in your booking platform, in Stripe, in your access control system and on your ClassPass ledger are four different records with four different IDs and four subtly different meanings. Ask a raw model a question across those silos and it joins them by guesswork. You get a confident answer built on invented joins, confidently wrong. Kula Intelligence exists for exactly this. Ingest the systems. Define one truthful model of the studio member, visit, plan, teacher, dollar. Apply it as a governed layer that any AI can safely use. ## The same truth, two different bets Here's where the paths split, and I think both choices are right for their markets. Lassie builds the substrate, the agent and the interface as one closed loop. They take a single job posting insurance payments, to near-total automation, then move to the next job. For US dental billing that's the correct call. The work is rule-bound, the payer policies are documented, correctness is everything, and the buyer wants to hand the job off entirely. The incumbent they replace isn't software. As Rampell put it, the incumbent is Betty, and she quit two weeks ago. Kula's bet is different: build the substrate, don't build the assistant. A studio owner connects their systems to Kula Intelligence and asks questions in the assistant they already use, Claude, ChatGPT or Gemini. Ask anything about your studio, grounded in your own data. Then act on it. It's self-serve at US$149 a month, and a single-site studio's data loads in about half an hour. Why the different bet? Because studio work is less rulebook and more judgement. Retention, pricing, the timetable, teacher relationships. The scarce resource in a studio isn't claim-posting labour, it's truthful understanding. So we start there, and automation grows outward from it as skills on the same substrate: the At-Risk screen, the Yield screen, and a growing set built by us and by others. Lassie automates a known job to free the human. We give the human leverage over the judgement calls first, and automate from there. ## The people who deliver the craft The line from the episode that travelled furthest came from Lassie's first customer: the agent isn't replacing humans, it's freeing them from wearing so many hats. The dentist got his evenings back and coaches his kids' soccer team now. In a dental practice, the owner and the practitioner are usually the same person, so freeing one human frees the craft. A studio isn't built that way. The owner carries the judgement calls, but the craft is delivered by a team of teachers and the deepest knowledge of your members lives in their heads, exactly the way Frédéric found the real billing workflows living in the heads of office managers rather than anywhere on the internet. So the substrate has two surfaces at Kula. The ask model is the operator's: the owner, in their own assistant, with the truth of the business behind every answer. Kula Tribe is the teachers': the same governed substrate, facing the people who stand in front of your members. Lassie frees the practitioner from admin. Tribe puts the substrate in the practitioners' hands because in this industry, teachers aren't overhead to automate around. They're the product. ## Underhyped where it matters The episode's most quoted line is that AI is overhyped in Silicon Valley and underhyped in Iowa. It's underhyped in Penrith and Paddington too. The hard part was never the model. It's reaching busy, non-technical owners with something that works in minutes, not months, which is why both companies obsess over consumer-grade onboarding, and why both were built by doing the work by hand first. Steijn spent months on a bar stool in a dental back office. Kula was built inside real studios. Nobody is against cleaning up busy work for people who'd rather be teaching, coaching or on the floor with members. That's the whole point. AI allows businesses to be human again. --- *Kula Intelligence connects your booking platform, payments and member systems into one governed layer you can ask anything in Claude, ChatGPT or Gemini. kula.digital* --- # Your business doesn't have an AI problem. It has a meaning problem. URL: https://kula.digital/blog/your-business-doesnt-have-an-ai-problem-it-has-a-meaning-problem Published: 2026-07-26 TL;DR: AI struggles with business data not because models are weak, but because data spread across disconnected systems has no agreed meaning. Kula Intelligence fixes this with three moves done once: an ingestor that normalises every source into one canonical stream, a machine-readable ontology map that defines what every entity and event means, and skills that turn the clean model into decisions. Because it ships as an MCP server, the whole team asks the same system and gets answers scoped to what each person is allowed to see: real data, or a plain statement that it can't answer and why. Point an AI at your business and ask it a simple question. Which members are at risk this month. How did Tuesday mornings do last quarter. You'll get an answer. It will sound confident. And some of the time it will be quietly, completely wrong. The usual response is to blame the model. Wait for a smarter one, write a longer prompt, add another tool. But the model was never the problem. ## The meaning problem A studio of any size runs on five or six systems. Bookings in one, payments in another, accounting in a third, marketing in a fourth. Every one of them has an API. Not one of them composes with the others. The same person exists in all of them: a member in the booking system, a customer in payments, a contact in marketing. Three records. Three ID schemes. No agreement anywhere about what a customer actually is. Hand that to an AI agent and it has to guess how the pieces join. It guesses mid-answer, silently, with no idea that it's guessing. The answer arrives fluent and plausible, built on a join that never existed. That's not an AI problem. It's a meaning problem. Nobody ever told the machine what anything means. ## Three moves, done once The fix is unglamorous, which is probably why it's rare. It's three moves, and the point is that you make them once. **Build the ingestor.** Every source comes in, API where one exists and CSV where one doesn't, and lands in a single canonical stream. A booking becomes a booking regardless of which platform it came from. The differences between systems get resolved at the door, not in every conversation afterwards. **Build the ontology map.** Every entity, relationship and event gets defined exactly once, in machine-readable form that travels with the data. What an active member is. What at-risk means. How attendance and payment state relate. When an agent asks "who's at risk this week," it isn't pattern-matching column names. It's reasoning against definitions. Two agents asking the same question get the same answer, because the meaning is in the data, not in the prompt. **Add the skills.** A clean model is necessary but it isn't the point. The point is the frameworks that run on top of it: the analysis nobody had time to do, the work that used to be a consulting engagement. An instructor bonus model that measures capacity against the true median class size for the time slot instead of how many bodies fit in the room. A first-thirty-days read that tells the team which new members need a person, not another automated message. ## A brain that knows its own edges Do those three things and the business has something new: a brain it can talk to. Ask it anything and one of two things happens. It hands back real data, with the reasoning visible. Or it says plainly that it can't answer, and why. The data isn't connected, the definition doesn't exist, the question is outside what it holds. I'd argue the second behaviour matters more than the first. Most AI in business today will answer anything you put to it. That isn't a feature. A system that knows the edge of its own knowledge is one you can act on. A system that guesses is a liability with a good interface. ## The shared part Here's the piece we didn't fully appreciate until we watched it in use. Because Kula Intelligence is built as an MCP server, the open standard Claude, ChatGPT and Gemini use to connect to real systems, it isn't one person's clever AI setup that nobody else can reproduce. It's shared infrastructure. The owner, the manager and the person on the front desk all ask the same system, against the same definitions. Each sees exactly the slice their role, their relationships and the member's consent allow. Every query passes through the same policy engine and lands in the same audit log with its declared purpose. Same language, different windows onto it. Nobody walks into a meeting with a different number for the same word. Getting a business to agree on what its own words mean turns out to be worth more than any single answer the system gives back. ## Where this is live The technical detail, including what sits behind the endpoint, the semantic catalogue, the five-layer policy engine, and what's connected today versus on the roadmap, is documented in full on the [Kula Intelligence MCP page](https://kula.digital/kula-intelligence/mcp). Connect it to the AI plan you already have and ask your own numbers a real question. And while fitness and wellness is where we built this, nothing about the method is fitness-specific. Ingest, define, apply. Any business whose data lives in systems that don't talk is the same shape of problem, and the same three moves solve it. --- # Chasing new clients is one strategy. Going back is the cheaper one. URL: https://kula.digital/blog/dont-chase-new-clients-go-back Published: 2026-07-21 TL;DR: Past clients are the cheapest, warmest growth a fitness consultant owns. AI finally gives you something new to bring them, so run one reactivation campaign across your old client book before the market catches on. Chasing new clients is a real strategy. Plenty of consultants build their whole practice on it. It just happens to be the most expensive and least certain thing you do, and most of us reach for it while a warmer option sits idle. Here is the number worth sitting with. The clients you have already worked with trust you more than any prospect ever will, and most consultants I know have stopped talking to almost all of them. You did the work. You fixed their bookings, or their pricing, or their retention. Then the engagement ended, they got busy, you got busy, and the relationship quietly went cold. Not because it failed. Because there was no obvious reason to call, and no easy way to bring them something genuinely new. That second part just changed. ## Why is your old client list your cheapest growth? The clients you have already helped are the lowest-cost, highest-trust growth available to you. The trust is already banked. The relationship history is already there. One consultant with a hundred past clients is sitting on a hundred warm conversations, each carrying years of context. For years, going back to an old client meant showing up empty-handed. "Just checking in" is not a reason to meet, and everyone on both sides of that call knows it. So consultants did the expensive thing instead. They chased strangers. Cold outreach, paid ads, referral begging, the whole tiring machine of winning trust you already had with the people you already helped. ## What can you bring them now that you couldn't before? What changed is not that your old clients suddenly want you back. It is that you now have something real to bring them. Over the last year, the tools available to a small fitness business have moved further than they did in the previous ten. A studio that could only ever see last month's numbers can now see what is happening this week: which members are quietly drifting before a cancellation lands, where money is leaking through double memberships and failed cards nobody chased. None of your old clients have this yet. Almost none of them know it exists. And you are the person they already trust to tell them. That is a reason to call. A real one. ## Why does reactivation beat acquisition on every number? New client acquisition is the most expensive thing you do, and the least certain. Reactivation is the cheapest and the most certain, because the trust and the context are already paid for. Think about what that does to your economics. A past client book is not a prospecting list. It is a pipeline you already paid for and forgot you owned. Every name on it carries years of context, and now every name carries a concrete new reason to talk. ## How do you actually run this? The move is not complicated. Take the clients you have worked with, going back as far as you like, one year, three years, five. Pick the ones whose businesses you understood well enough to genuinely help again. Then go back to them, not with "how are things," but with "the ground has shifted since we worked together, and there is something here that would change how you run your studio. Let me show you." Run that as one campaign across the whole list. You will get more meetings from it than from a month of chasing new logos, and you will get them faster. ## Where I sit in this I am biased about the delivery end of this, so let me be plain about it. I run a studio myself, and I build the intelligence layer that produces exactly the kind of insight that gives a consultant something new to walk back in with. If you want to be the person who brings that to your client book rather than watching someone else do it, that is a conversation worth having. But the bigger point stands on its own, whether you build it, buy it, or partner on it. The clients you have already helped are the least contested, lowest-cost, highest-trust growth you will ever find. For the first time, you have a reason to go back to all of them at once. Most consultants will keep chasing strangers. The ones who go back first will own the relationships before the market wakes up. --- # Yoga and boutique fitness in 2026: Australia, the UK and the US, and the one thing they all get wrong URL: https://kula.digital/blog/yoga-boutique-fitness-2026-au-uk-us Published: 2026-07-21 TL;DR: Across Australia, the UK and the US, boutique fitness is at three different stages of one story. The winners in every market are converging on the same two problems: retention, which is really a visibility problem, and instruction, which is really a development problem. Neither is marketing. If you run a studio, it is worth occasionally lifting your head from your own timetable and looking at the market you are actually operating in. Right now, three of the most mature English-speaking fitness markets, Australia, the United Kingdom and the United States, are telling three versions of the same story. Read together, they point at where this is all heading, and at two problems the winners in every market are quietly converging on. Start with the shape of the whole thing. The global boutique fitness market is now valued at roughly sixty billion US dollars and growing at a high single-digit rate, depending on whose report you read. The narrower category of yoga and Pilates studios is one of the more resilient corners of it, forecast across the major research houses to keep compounding at double-digit annual rates into the 2030s. So the category is healthy. But the headline growth hides very different weather in each market. ## Australia: mature, plateauing, and over-spending on the wrong thing Australia is the most instructive of the three, because it is the furthest into the plateau. IBISWorld puts the gym and fitness centre sector at around three point seven billion dollars in 2025, and notes it actually dipped slightly that year. There are somewhere around seven and a half thousand gyms and clubs operating nationally, and boutique studio revenue is estimated to have peaked back in 2023. In other words, the easy growth is behind us. The market is full. When a market is full, the game stops being about finding new members and starts being about keeping the ones you have. And this is where Australian operators, on the whole, are getting it wrong. The boutique platform bsport, in its 2026 studio playbook, made the point bluntly: acquiring a new studio member costs somewhere between five and seven times more than keeping an existing one, and yet the average Australian boutique studio still spends far more on acquisition and marketing than it ever spends on retention. Sit with that. The most expensive, least certain path to growth is where the money goes, while the cheapest, most certain path is treated as an afterthought. In a plateauing market that is not a small inefficiency. It is the whole ballgame. ## The United Kingdom: still climbing, and splitting in two The UK is a few steps behind Australia on the curve, and still growing. IMARC has the UK fitness and gym market at around seven point four billion US dollars in 2025, with membership up six per cent the year before to roughly eleven and a half million people. Healthy on the surface. Underneath, the more interesting story is structural. Independent analysts describe a market that is polarising hard, splitting into a budget pole and a premium boutique pole, with the traditional mid-market getting squeezed out between them. The boutique end is winning on experience, instruction and community rather than price. And within boutique, Reformer Pilates has emerged as the standout format, described by UK market watchers as one of the fastest-growing and most structurally resilient categories, precisely because it combines a real physical result with a sticky, community-led studio experience. The other signal worth noting: the boutique wave, which started in London and a handful of big cities, is now spreading into secondary cities, commuter towns and affluent suburbs. If you are a UK operator outside the capital, that is your window opening. ## The United States: the most mature, and rotating The US is the oldest and largest of the three markets, and it shows. It is home to the big branded boutique machines, the Barry's, SoulCycles, CorePowers and Solidcores of the world, operating at a scale the other two markets have not reached. But maturity brings rotation. Search-interest data suggests American appetite for the word yoga specifically has softened over the last several years, even as Pilates and broader boutique formats surge. That does not mean yoga is dying. It means the category is maturing and the demand is moving, from yoga as the headline act toward Pilates, strength and hybrid formats, with yoga increasingly woven in as one thread rather than the whole cloth. Studios that built their entire identity on a single format are the ones most exposed to that rotation. ## The two problems every market shares Put the three side by side and the differences are really just timing. Australia is showing the UK and US their near future: a full market where retention decides everything. And across all three, underneath the local detail, the same two problems keep surfacing. The first is retention, and the strange refusal to fund it. Every market over-invests in acquisition relative to keeping members, despite the maths being wildly in retention's favour. This is the single most consistent, most expensive mistake in the global studio business, and it is almost entirely a visibility problem. Studios spend on acquisition because acquisition is measurable and visible. They under-spend on retention because most studios genuinely cannot see who is about to leave until they have already gone. The second is instruction, and the supply of it. Boutique wins on the quality of the teaching in the room. That is the entire value proposition. Yet the global picture shows a highly fragmented instructor base, with the large majority of yoga studios worldwide running on fewer than five instructors. The people in the room are the product, and they are also the constraint. A studio is only ever as good as the teachers it can attract, develop and keep, which makes the training and development of instructors not a nice-to-have but a core piece of competitive infrastructure. ## What it means if you operate The lesson from reading all three markets at once is that the next phase of this industry will not be won the way the last one was. The last phase rewarded whoever could open fastest and market hardest. The full, plateauing market that Australia is already in, and that the UK and US are heading toward, rewards something quieter and harder: keeping the members you have, and fielding teachers good enough that they want to stay. Both of those are about depth, not reach. One is a retention problem, which is really a visibility problem, being able to see a member cooling in time to do something human about it. The other is a talent problem, which is really a development problem, building and holding a bench of teachers who make people want to come back. The studios that treat those two as their real strategy, rather than as overheads to minimise while they chase the next campaign, are the ones that will still be standing when the market they are in finally fills up. In Australia, it already has. --- # Nobody's watching the gaps in your studio, and that's where the money goes URL: https://kula.digital/blog/money-slipping-out-of-your-studio Published: 2026-07-21 TL;DR: Your booking system and payment processor each do their own job and can't see across each other, so a few students slip through the gap between them. The most common leak is repeat intro passes: a light sampler re-buying your trial past the one-per-customer limit, which cost one Bondi studio around $13,000 of access below cost in a year. Caught on day one it's a no-lose bet, convert them or you have still doubled that student's value. Closing these gaps needs an independent layer that reads across all your systems, not another system with its own blind spots. Every studio has money slipping out the back, and most owners never see it leave. Not because they are careless. Because the systems they run were never built to show them. Your booking software takes bookings. Your payment processor moves money. Each does its own job well, and neither one is watching the other. The gap between them is where a handful of students quietly slip through, and it stays invisible for a simple reason: no single system can see the whole person. ## Why can't your own systems catch this? Here is the uncomfortable truth about the software running your studio. It was built to do a job, not to audit itself. Your booking system checks one account at a time, one email, one phone, one profile. Your payment tool sees transactions, not the human behind them. Neither was designed to raise its hand and say "this looks like the same person as that." Not because the products are bad, but because catching leaks was never what they were sold to do. No system is built to expose its own blind spots. That is not a knock on your software, it is just what it is for. Which is exactly why the gaps go unwatched. Nobody is checking the space between your systems, because none of your systems can. ## The big one: repeat intro passes A repeat intro pass user is a student who buys your trial a second or third time instead of joining, using a fresh email, phone, or card each time to slip past your one-per-customer limit. Same person, new account, and nothing in a standard database connects the two. This one is common and known across the industry. Legacy booking systems check email or phone, and modern payment habits walk straight through. People do what a system lets them do. Be honest about who this is. One to three intro passes is not loyalty. It is a light sampler who liked the place enough to come back. Thin, but it is the useful half of a sale already done: they have shown they want to be in the room. The only thing left is to move them onto the plan that fits how they practise while a pass is live. ## What does it cost? At one Bondi studio, over 12 months, 72 students bought a repeat trial. Between them they took 905 classes on those recycled passes and paid about $6,695 in trial fees. Priced as membership-equivalent access, that is roughly $13,000 of unlimited access handed over below cost in a year. One honest caveat so you can trust the number. That $13,000 is not cash gone from the till. It is access given away cheap, and some of those students would have walked rather than pay for a membership, so not every dollar was recoverable. But even half is real, and the forward-looking prize is bigger. ## Why the old way of handling it backfires The usual move is to catch one by fluke, dig through records to confirm it, then ring them up annoyed. That is the worst option available. The moment you name the second pass, they feel caught. Someone who feels caught does not upgrade. They book less and drift off, and you have turned a paying visitor into an empty spot. You were never losing here. Someone practising at your studio and paying to do it is not a loss, whatever channel they came through. ## The fix: treat it as the wrong plan, not a crime Stop treating it as enforcement. Treat it as someone on a pass that no longer fits. Catch it on day one of the second pass, not near expiry, and send a warm note that never mentions the pass: > Hi [name], it's Bec from the studio. I've loved seeing you on the mat lately, you've built a really nice rhythm. I'd like to get you onto the plan that fits how you're training, so there are no gaps between passes and you get priority booking. I can set you up on the fortnightly membership with a welcome offer on your first month. Want me to sort it? Sell it on convenience and belonging, not price. At a high practice rate a membership is not cheaper per class than a recycled trial, and this student is sharp enough to know it. What a membership gives them is no gaps between passes, no re-buying admin every three weeks, priority booking, and arriving as a member rather than a guest. That ask is honest, which is the whole reason it works. ## Why this is a bet you can't lose Catch them on day one of that second pass and they have already paid you for it, say $49. From there, two things can happen. They upgrade to a membership. That is the real win: recurring revenue from someone who has already shown they will come back and pay. Or they do not, and they drift off quietly. Even then you have banked two passes instead of one and doubled what a single-trial student was ever worth to you, without making anyone feel caught. Because this is a light sampler and not a treasured regular, the downside of reaching out is close to nothing. There is no version of this where you come out behind. ## The rarer one, which I won't detail There is a second leak I found in the same studio, and I am not going to explain how it works, because a how-to is the last thing this problem needs. In short, a small number of members lean on the billing grace period, the buffer that keeps you booking when a payment fails, to keep practising while paying less than they owe. It is rare, and unlike the intro pass it is deliberate. The only reason I mention it is this: the studio had no idea it was happening. It showed up nowhere they would look. It surfaced because something was finally reading across the booking record and the payment behaviour at once, instead of one system at a time. ## The real point: someone has to watch the gaps Your booking software cannot audit itself. Your payment processor sees the money move but not the person moving it. Even a great payments tool like Stripe only knows what a card did, not who was holding it or which second account it belonged to. Each system is honest about its own patch and blind to the space between them. That gap is where the leaks live, and closing it needs something none of those systems can be: independent. A layer whose only job is to read across all of them and check what no single one can see on its own. That is what Kula Intelligence is for. It sits above your booking and payment systems, reads across the accounts, and names the pattern, matched on the signal hardest to fake, the same card behind a different name or email. For the intro pass, it flags day one and drafts the message, so the manager sends a warm invite in one tap instead of digging through records for an hour. It is not another system doing its own job. It is the one checking all the others. You cannot close a leak you cannot see. The first step is having something in place whose whole purpose is to see it. --- # The audit you never ordered, already ran URL: https://kula.digital/blog/the-audit-you-never-ordered-already-ran Published: 2026-07-21 TL;DR: A data lake brings a studio's records together, but records find nothing on their own. An ontology — the map of what things are and how they relate — turns linked data into findings: fraud hiding in 0.01% of activity (a member re-signing up with a different email address but the same credit card to cycle discounted intro passes), an intro plan capped at five visits when the habit that keeps members forms at six to ten visits in 28 days, and data gaps that silently overstate revenue. Kula Intelligence builds the fitness ontology once and runs it for every studio, and the value compounds with every location. The short version Last time I wrote about the lake — the place all of a studio's history finally lives together. This post is about the layer on top: the ontology, the map that tells the system what every record *is* and how it connects to everything else. Palantir and Databricks built enterprise empires on this idea. We've brought it to boutique fitness. The outcome isn't an architecture diagram — it's found money: fraud hiding in 0.01% of activity, an intro plan capped one visit short of the habit threshold, and data gaps that silently overstate revenue. And every one of those findings compounds when you run more than one location. - A data lake gathers your studio's records into one place, but a pile of connected records finds nothing on its own. Meaning requires a map. - An ontology is that map: a member is a person, a visit belongs to a member, draws on a plan, was paid for by a sale, taught by a teacher, at a location. - Once every record has a place on the map, the map tells you what *should* exist — and that's when the money starts turning up. - The anomalies that matter don't live in any single system. They live in the relationships between systems, which is exactly what an ontology holds. - In a studio, leaked revenue comes straight off the profit line, because the class ran and the teacher was paid regardless. Small leaks hit profit hard. - The benefit multiplies with locations: one map, applied everywhere, makes drift comparable across sites — one question instead of five audits. In [the last post](https://kula.digital/blog/the-analyst-you-cant-afford-already-hired) I made the case for the lake: bring all of a studio's history into one connected place, the way enterprises do, and run it for operators who could never staff it. That post ended with the linking — every visit tied to the plan it drew on and the sale that paid for it. This post is about what the linking is *for*. Because here's the uncomfortable truth about data lakes: on their own, they find nothing. A lake is a very well-organised pile. The layer that turns the pile into answers has a name the enterprise world uses and nobody else loves. ## An ugly word for a simple idea **Ontology.** It sounds like a philosophy seminar. It means something a studio owner already understands: an agreed map of what things are and how they relate. In a fitness studio, the map reads like this. A member is a person. A visit belongs to a member. The visit draws on a plan. The plan was paid for by a sale. The class was taught by a teacher, at a location, in a room, at a time. Every record that lands in the lake gets a place on that map — not just stored, but *understood*. If this sounds abstract, look at who's built businesses on it. Palantir's moat has never been its models — it's the ontology its engineers spend months building inside each enterprise customer, at seven-figure prices. Databricks spent 2026 proving the same point from the other direction: AI agents fail on context, not intelligence, and the context layer is the fix. Y Combinator went as far as naming the "Company Brain" a Request for Startups. The pattern is validated at the very top of the market. It has simply never been available at the bottom — for the same reason the lake wasn't. It takes a team to build, per customer, and a boutique studio can't hire that team. Our answer was the vertical: every boutique fitness studio shares the same primitives, so we built the fitness ontology once, and every studio that connects gets it on day one. So what does it actually *do*? Here's the key sentence of this whole post: **once every record has a place on the map, the map tells you what should exist.** A visit should have a plan behind it. A plan should have a sale behind it. A sale should match a price on the list. A refund table at a busy studio should have rows in it. When reality doesn't match the map, that mismatch is a finding — and findings, it turns out, are frequently money. Three stories. ## The fraud: 0.01% of activity, none of it visible At a multi-location operator we work with, the map surfaced a member who kept becoming a new member. A fresh sign-up, a slightly different email address every time — and the same credit card underneath. To the booking platform, each sign-up was a brand-new person, entitled to a brand-new, heavily discounted intro pass. To the payment data, it was one person on a loop, paying the new-member price forever instead of ever becoming a member. Neither system could catch it alone — and this is the detail worth sitting with. The booking platform saw a stream of legitimate new faces; every record was valid. The payment platform saw routine charges on an ordinary card; every transaction was clean. The pattern only exists when identity is resolved *across* the two — when the map insists that a person is a person, however many email addresses they wear, and the credit card becomes the thread that ties the masks together. (It wasn't the only pattern, either. A second one involved gaming the grace period between payments — that one deserves, and will get, a post of its own.) Call it 0.01% of activity. That sounds like nothing. Studio economics say otherwise. Revenue that should have arrived and didn't comes off the bottom line dollar for dollar, because every cost was already sunk — the class ran anyway, the teacher was paid anyway, the lights were on anyway. At typical studio margins, a fraction of a percent of revenue leakage translates to a multiple of that off profit. The 0.01% punches far above its weight. > Nobody ordered an audit. The data just finally knew what it was supposed to look like. Relationships are exactly what an ontology holds. That's the whole trick. ## The cap nobody changed back The second finding wasn't fraud. Nobody did anything wrong — which is exactly what makes this the more expensive category. The operator's intro plan — the discounted pass that gives a new member their first taste of the studio — had been capped at a maximum of five visits. There was a good reason at the time: classes were running at high utilisation, and gating intro visits protected capacity for paying members. Then more classes opened, utilisation eased, and the cap stayed. Settings don't announce themselves. Nobody changed it back, because nothing looked wrong. Here's why it mattered. Behavioural analysis of more than a thousand intro buyers at this operator shows a clear habit threshold: the members who stay are the ones who get to six to ten visits in their first 28 days. Below that, the habit doesn't form and the conversion doesn't happen. The intro plan was capped at five — one visit short of the bottom of that range. The product designed to create members was configured to stop people just before the point where they become one. No single system could see this either. The plan setting lived in the booking platform. The habit threshold lived in attendance behaviour. The cost lived in conversion outcomes. Three different places, one finding — and it only exists when plan configuration, visit behaviour and conversion sit on the same map. A revenue report shows the intro pass earning nicely. Only the map shows what it's quietly preventing. ## The gaps that read as zero The third category is my favourite, because it's invisible by definition. At one studio processing six figures a month, we found the refund tables completely empty. Not because there were no refunds — a business at that volume certainly issues some — but because refund events weren't flowing from the booking platform at all. Without a map, an empty refund table looks like good news. With one, it's a finding: the map says refunds *should* exist here, so their absence means net revenue is silently overstated, and every downstream number that depends on it inherits the error. Same logic everywhere the map has expectations. A month with zero attendance at a studio that was open is a data gap, not a quiet month. A member with visits but no plan isn't a loyal ghost — it's a broken link worth chasing. An ontology is the only thing that makes *absence* visible, because absence is only detectable against a statement of what should be there. ## Multiply by locations For a single studio, everything above amounts to a very good audit that nobody had to commission. For a multi-location operator, the economics change shape. The same map applies at every site. That makes drift *comparable*: "which location's pricing has moved furthest from the list?" is one question with a ranked answer, not five separate audits with five separate consultants. A finding at one site instantly becomes a question for every site — is any other location still carrying that intro cap from its own busy season? Leakage that rounds to nothing at one site compounds across ten. And the anomalies cluster — the map doesn't just find the leak, it shows you where the leaks concentrate, which is usually where the process is broken, which is the thing actually worth fixing. This is also why the enterprise world paid so much for the pattern for so long. The value of an ontology scales with the surface area it covers. Studios were locked out not because the value wasn't there, but because the build cost was per-customer. Building it once for a vertical is what breaks that. ## What it means The lake was the plumbing. The ontology is the meaning. Together they're the thing YC is calling the Company Brain — the layer Databricks and Palantir proved at enterprise scale, running for a business that could never have bought it. And the outcome, for an operator, isn't a report. It's that the audit never stops. Every night the data lands, takes its place on the map, and either matches what should exist or doesn't. The fraud, the misconfiguration, the gaps — none of them were found because someone went looking. They were found because, for the first time, something was *always* looking. Studios are not short on dashboards. They're short on answers. It turns out some of the most valuable answers are to questions nobody thought to ask. --- # You get paid for leads. You get judged on members. URL: https://kula.digital/blog/fitness-studio-retention-agencies Published: 2026-07-21 TL;DR: Fitness studios hire agencies on cost per lead and fire them over retention. The members leave, the owner blames the agency, and the fix is not more ads. It is giving instructors one piece of information before class: who has not been in for a while. If you run marketing for fitness studios, you already know the trap even if you have never named it. You are hired on cost per lead. You are measured on cost per trial. You report on both, and the numbers are usually good, because generating interest in a boutique studio is not the hard part. Then, somewhere around month four, the tone changes. The owner is less responsive. The retainer gets questioned. Eventually you are cut, and the reason given is vague. Not "your ads failed." Something softer and much harder to argue with. We just weren't seeing the value. Here is what happened. The members you brought in stopped showing up, and the owner counted that against you. ## The problem you are being blamed for is not yours The economics of a boutique studio are brutal in a way that is invisible from the outside. Members leave at a rate that would be considered a crisis in any other subscription business. Teachers leave at a similar rate, and when a teacher goes, a portion of their members quietly go with them. Faced with a leaking bucket, most owners do the only thing they know how to do. They pour more in. That means more acquisition spend, which means you. So the agency gets hired to compensate for a retention problem, and then gets held responsible when the retention problem continues. You are being asked to solve churn with reach, and it does not work, and everyone involved slowly concludes it is your fault. ## Why every fix has failed Ask any studio owner what they have tried on retention and you will hear the same list. Loyalty programmes. Challenge months. Win-back emails. Referral bonuses. And, at some point, an instructor incentive scheme. That last one is the interesting failure, because it is aimed at exactly the right place and it still does not work. The logic is sound. Instructors have the relationship. Members do not stay for a studio, they stay for a teacher and a room and a time slot that fits their week. If you want to influence retention, that is where the leverage is. So owners try to activate it, and they do it the only way they know: by asking instructors to sell. Push the ten-pack. Mention the challenge. Upsell to unlimited. It fails universally, and the reason is not laziness or lack of incentive. It is that nobody becomes a Pilates or yoga teacher to sell things. Ask them to and you are asking them to become someone they specifically chose not to be. They will comply badly for a month and then stop. ## The reframe There is something instructors want very much, and it costs nothing to give them. They want to be the teacher who notices. Every good instructor already carries a rough version of this in their head. They know roughly who has been missing. They have a vague sense that someone has dropped off. But they teach across multiple classes, sometimes multiple sites, and their memory is doing a job that no memory can do reliably. So the intervention is not an incentive. It is information. Tell a teacher, before class, that one of their regulars has not been in for three weeks. Do not give them a script. Do not give them an offer to make. Do not give them a target. Just give them the fact, and get out of the way. What happens next is the whole product. They say something at the door. "Haven't seen you, everything alright?" Five words, no commercial content, and a member who was drifting is now a member who was noticed. That is not selling. It is the thing the instructor already wanted to be good at, and the only reason they were not doing it is that nobody handed them the information. ## What changes for you If retention lifts, three things happen to your agency and all of them are good. The account survives longer, because the owner can afford you and can see the value. Your attribution extends past the intro pass, so you can report on members retained rather than leads generated. And you stop competing on cost per lead against every other agency in the market, because you are the only one who can talk about what happened after the sale. That last one is the real prize. Cost per lead is a race to the bottom and you already know it. Members retained is a completely different conversation, and it changes what you are worth. ## The uncomfortable part None of this works without data the studio owns and most owners have never looked at. Attendance patterns. Which teacher each member is actually bonded to. How long the gap has been. Whether they are still being billed while not showing up. It exists. It sits in the booking system. It has never been turned into anything an instructor could use ninety seconds before class. That is the gap. Not more leads. Not better ads. Just telling the person who already has the relationship what they need to know to use it. --- # From three years of expensive lessons to day one URL: https://kula.digital/blog/expensive-lessons-to-day-one Published: 2026-06-29 TL;DR: Opening a studio means closing a knowledge gap that used to take three years of expensive lessons. A Gen 2 studio gives a first-time owner that help on day one, and keeps the artist as the artist. The barrier to opening a studio was never nerve, and never the yoga. It's a knowledge gap. The good news is the gap closes on day one now, before it costs you anything. ## What actually stops a great teacher running a great business? Capital, leases, contracts, the operational machinery, none of it is on a teacher training. You can't search your way out because you don't yet know the questions to ask. So most owners learn it the slow way, over about three years of expensive lessons. ## What does a first-time owner hold on day one with a Gen 2 studio? The help that used to take years to assemble, available from the first morning. The read on what's really happening in the business, the nudge when a member starts to drift, the three things worth doing today, the cover filled, the marketing read back plainly, and a straight answer whenever "I don't know what I don't know" comes up. Read that list back as a job ad. A new owner could never hire half of it. A Gen 2 studio closes the gap before it costs a thing. ## Does this replace learning the craft? No, it speeds it up and keeps you on the rails while you learn. The system can flag when a decision is about to cost more than it saves, which is how a first-time operator stays inside good practice from day one instead of after three years of finding out the hard way. That's not a constraint. It's a head start. --- This is part of the Gen 2 studio series. Start with the pillar: [The Gen 2 studio: when every decision finally has help behind it](/blog/gen-2-studio-help-on-every-decision). See also [The team a single studio could never hire](/blog/team-a-single-studio-could-never-hire). --- # Behind the desk, not in the room URL: https://kula.digital/blog/behind-the-desk-not-in-the-room Published: 2026-06-29 TL;DR: AI does not hollow out a studio; drudgery does. A Gen 2 studio runs the back-office watching and hands the human moments to the team. Intelligence behind the desk, humans on the floor. The fear is that AI hollows out the human thing a studio is built on. Built right, it does the opposite. It clears out the machine work so the people have room to be people. ## Doesn't AI make a studio less personal? What makes a studio feel like a machine today isn't the people. It's the drudgery crowding them out: the spreadsheet that won't talk to the other spreadsheet, the 6am cover scramble, the Sunday night on the laptop while the family waits. ## Where does the intelligence belong? Behind the desk, never in the room. Kula Intelligence takes the watching, matching and chasing and leaves the relating to your team. It never speaks to a member like a robot. It speaks to your people and hands a teacher the one true thing they need, "Sarah's back after three weeks, welcome her home," then gets out of the way. ## What does the evidence say members actually want? In Les Mills' 2026 research, only about one in ten members would choose AI guidance over a human coach. People are happy to let software run the back office. They want the room kept human. So that's exactly where the line sits: intelligence behind the desk, humans on the floor. --- This is part of the Gen 2 studio series. Start with the pillar: [The Gen 2 studio: when every decision finally has help behind it](/blog/gen-2-studio-help-on-every-decision). Next, read [From three years of expensive lessons to day one](/blog/expensive-lessons-to-day-one). --- # Forty decisions, time for three URL: https://kula.digital/blog/forty-decisions-time-for-three Published: 2026-06-29 TL;DR: Owners make forty decisions a week with time for three. A Gen 2 studio ranks them by what is worth most and what will not wait, so you act on the right three instead of guessing. The most important thing you could do today and the thing that feels most urgent are almost never the same. The gap between them is where the easy wins have been hiding all along. ## Why is choosing the right three the hard part? Forty things could be done today. There's time for three. The squeaky problem grabs the morning, the quietly valuable one waits, and over a year that pattern is where a lot of upside slips by unclaimed. ## How does a Gen 2 studio decide what comes first? It prices the choices. It knows this member has been a regular for two years and is a week into the kind of drift that's simple to turn around early, while that one is just on holiday and will be back. It knows the Tuesday 6am is sitting below its potential and the Saturday class has a waitlist worth using. So instead of forty undifferentiated tasks, you get a short, ranked list: do this first, it's worth the most and it won't wait; leave that, it can hold. ## What changes when the ranking is done for you? You stop guessing. The same hours go to the three calls that move the studio instead of the three that shouted loudest. That's a straight answer on what to do first, which is the help an owner has never actually had. --- This is part of the Gen 2 studio series. Start with the pillar: [The Gen 2 studio: when every decision finally has help behind it](/blog/gen-2-studio-help-on-every-decision). See also [The team a single studio could never hire](/blog/team-a-single-studio-could-never-hire). --- # The team a single studio could never hire URL: https://kula.digital/blog/team-a-single-studio-could-never-hire Published: 2026-06-29 TL;DR: Big chains run on a back-office team watching each slice of the business. A single studio could never afford one. A Gen 2 studio plays all those roles at once for one owner, behind the front desk, for less than one membership. A big chain has a person whose entire job is to notice when a member starts drifting. You have a front desk, a timetable to teach, and a memory that's already full. That was never unfair. It was just unaffordable. Until now. ## What does a studio actually need behind the front desk? Picture the team a large group takes for granted. One person watching every member against their own normal. One reading what each class is really doing. One who fills a cover before it becomes a scramble. One who knows which marketing brought people through the door. Four sets of eyes, each on one slice, each catching what the owner can't get to. A single-site owner has none of them, so the owner is all four, after teaching, on a Sunday night. ## Why couldn't a small studio ever hire this team? Because the maths never worked. Read the four roles back as a job ad and the salary is more than most studios turn over. So the work doesn't disappear, it just lands on one person who's already at capacity, and the slices nobody has time to watch are exactly where the quiet wins sit. ## How does a Gen 2 studio give one owner that team? A Gen 2 studio runs the watching for you. Kula Intelligence reads the whole business as one connected picture and plays every one of those roles at once, behind the front desk, for less than the price of one membership. It knows what each class is doing for you, not just how full it looks. It notices when a three-times-a-week regular eases off to once, while there's still time to say something warm. It hands a teacher the cover before you've reached for your phone. It reads which of your marketing actually worked. Same owner, same hours, a whole team's worth of attention finally pointed at the business. --- This is part of the Gen 2 studio series. Start with the pillar: [The Gen 2 studio: when every decision finally has help behind it](/blog/gen-2-studio-help-on-every-decision). Next, read [Forty decisions, time for three](/blog/forty-decisions-time-for-three). --- # The Gen 2 studio: when every decision finally has help behind it URL: https://kula.digital/blog/gen-2-studio-help-on-every-decision Published: 2026-06-29 TL;DR: Studio software used to just record what happened. A Gen 2 studio reads the whole business, helps you decide what to do next, and hands a single owner the kind of expert help only big chains could afford, on every decision, behind the front desk while the humans stay on the floor. Run a studio and you make about forty decisions a week. You get to properly think through maybe three of them. The rest happen on the fly, between classes, shaped by whatever felt loudest that morning. That was never a talent problem. It was a help problem. And it's the thing that just changed. ## What's the difference between a Gen 1 and a Gen 2 studio? A Gen 1 studio runs on software that records what already happened. A Gen 2 studio runs on software that helps you decide what to do next, and quietly handles a chunk of it for you. Gen 1 is a filing cabinet with a search box. It logs the class that ran half full, the payment that bounced, the member who cancelled. All true, all useful, all arriving just after the moment you could have used it. You see a regular has gone the week they go, not the month they first started drifting, which was the month a quiet word would have landed. Gen 2 reads the whole studio as one picture and reads it back to you in plain English. It catches the drift while it's still just drift and easy to turn around. It tells you what each choice in front of you is actually worth. And the part that matters most: it puts experienced help behind every decision you make. ## Why recording your business was never the same as running it For twenty years, software could store your studio but it couldn't understand it. It could draw you a chart. It couldn't tell you what the chart meant or what to do about it. So every operator ended up doing the reading themselves, late, on a laptop, after a full day of teaching. A report you can't act on in time isn't help. It's homework. And the calls that shape your studio deserve better than homework done at the end of a long day. ## What does help on every decision actually look like? Here's what a big chain always had and a single-site owner never could: a room full of people whose whole job was to watch one slice of the business and catch what the owner couldn't get to. A Gen 2 studio gives a single owner that team. Not as job titles. As help, on tap, on every call. Someone who knows what each class is really doing for you, not just how full it looks. Someone who notices when a three-times-a-week regular eases off to once, while there's still time to say something warm. The hand that fills a cover before you've reached for your phone. Someone who reads which of your marketing actually brought people through the door. Read that back as a job ad. A small studio could never hire half of it. Kula Intelligence is all of it at once, sitting behind the front desk, for less than the price of one membership. ## How does a Gen 2 studio know which decision matters most today? This is the part owners feel first. Forty things could be done today. There's time for three. Which three? The most important thing and the most urgent-feeling thing are rarely the same, and closing that gap is where the easy wins have been hiding all along. The squeaky problem gets the morning. The one worth the most quietly waits. A Gen 2 studio sorts it for you. It knows this member has been a regular for two years and is a week into the kind of quiet drift that's simple to turn around early, while that one is just on holiday and will be back. It knows the Tuesday 6am is sitting below its potential and the Saturday class has a waitlist worth doing something with. So instead of forty undifferentiated tasks, you get a short, ranked list. Do this first, it's worth the most and it won't wait. Leave that, it can hold. That's the help an owner never had. Not more to do. A straight answer on what to do first. ## Doesn't this make a studio less human? Fair question, and it's the one I care about most. Pour AI into a business built on human connection and surely you hollow out the thing that made it special. It does the opposite, if it's built right. What makes a studio feel like a machine today isn't the people. It's the drudgery that crowds them out. The spreadsheet that won't talk to the other spreadsheet. The 6am cover scramble. The Sunday night on the laptop while the family waits. Kula Intelligence takes the machine work, the watching and matching and chasing, and leaves the relating to your team. It never speaks to your member like a robot. It speaks to your people, hands a teacher the one true thing they need, "Sarah's back after three weeks, welcome her home," and gets out of the way. In Les Mills' 2026 research, only about one in ten members would choose AI guidance over a human coach. People are happy to let software run the back office. They want the room kept human. So that's exactly where we keep it. Intelligence behind the desk, humans on the floor. ## The proof, not the promise None of this is theory. At Body Mind Life in Bondi, one workflow alone, the friction-free membership pause, has run more than 3,000 times on its own over 18 months and handed back close to 600 hours of front-desk time. The owner didn't have to think about a single one of them. That's the shape of a Gen 2 studio. The same owner, the same hours, a completely different trajectory, because the help finally showed up where the decisions get made. A few years from now the best operators will all run this way, for the same reason nobody runs a business off paper ledgers any more. Once you've seen your whole studio clearly, with help on every call, you build from there. --- # The analyst you can't afford, already hired URL: https://kula.digital/blog/the-analyst-you-cant-afford-already-hired Published: 2026-06-28 TL;DR: There are two ways to let an AI read your studio's data: query each live system one question at a time (the API approach, like drinking through a straw), or bring all your history into one connected place the AI can read at speed (a data lake). Lakes win on speed, on connecting systems that don't talk to each other, and on safety, but they take a team of data engineers to run, which is why they've always been enterprise-only. Kula Intelligence builds and runs the lake for you, so the AI you already use can answer questions about your studio as if it had read every number in the building. When people ask how Kula Intelligence works, they usually expect me to name a model. Which AI is it? The honest answer is that the model is the least interesting part, and it's not the part we built. The work we did sits underneath the AI, in the boring plumbing nobody wants to talk about and everybody quietly needs. So let me talk about it. Because the choice we made there is the whole reason Kula gives you a straight answer about your studio in seconds, instead of a slow, half-right one. ## Two ways to let an AI read your business There are really only two ways to connect an AI to your studio's data, and the difference between them decides everything about what you get back. The first way is the one most people reach for. Leave your data where it is, in Mindbody, in Stripe, in your inbox, and have the AI ask each system a question every time you do. You want to know who hasn't been in lately, so the AI calls your booking system, waits, gets a slice of rows back, and reads them to you. This is the API approach. Think of it as drinking through a straw. One question, one thin tube poked into a live system, a few sips pulled back out. For a single lookup, a straw is fine. "When did Sarah last book?" One sip, done. But your real questions aren't single lookups. "How's the reformer cohort tracking over the last six months, and which of them are slipping?" is not one question. It's hundreds. Every member, every visit, every plan, every payment, cross-referenced across six months and three different systems that don't know each other exist. Through a straw, that's hundreds of sips, each one waiting in line behind the last. By the time the answer arrives (if it arrives), you've made a cup of tea and forgotten what you asked. The second way is the one we chose. Bring the data together first, into one place built for the AI to read fast, and let it wade through the whole picture at once. Not a straw into a live system. A lake the AI can swim in. ## Why the lake wins When all of your studio's history lives in one place, already pulled together, already linked, already cleaned up, three things change. It's fast. The AI isn't waiting on five different systems to answer one at a time. The whole picture is right there, so "six months of the reformer cohort" comes back in the time a straw would still be on its third sip. It's connected. This is the part that matters most and gets talked about least. Your booking system knows about visits. Your payment system knows about money. Neither of them knows the visit and the payment are about the same person, that link doesn't exist until someone makes it. In the lake, we make it. Every visit is tied to the plan it was drawn against and the sale that paid for it, so a question like "are the members on class packs staying as well as the members on memberships?" actually has an answer. Through a straw, it never could, because the straw can only ever see one system at a time. It doesn't break anything. The lake is a read-only copy. The AI reads your history; it never touches the live systems your studio runs on. It can read everything and break nothing. This isn't a new idea, by the way. The biggest companies in the world (banks, airlines, retailers) have been building exactly this kind of data lake for years. There's a whole industry around it. Databricks, the best-known name in the space, is built on the lake model and worth tens of billions of dollars. ## So why doesn't every studio have one? Because building and running one is a real job. Several jobs, actually. The companies that run data lakes employ teams to do it. Data engineers to pull everything together. Analysts to make sense of it. People whose entire week is keeping the plumbing clean so the answers stay trustworthy. That's why the lake has always been enterprise-only, not because small businesses don't need it, but because small businesses can't staff it. A boutique studio can't hire a data engineer. You shouldn't have to know what one is. You're already on the laptop on a Sunday night with five tabs open and spreadsheets that don't talk to each other, the last thing you need is a sixth system and a hiring problem on top of it. That's the gap Kula closes. We built the lake, and we run it for you. The pulling-together, the linking, the cleaning, the keeping-it-honest: the work an enterprise pays a team to do, happening underneath, handled, so you never see it. What you see is the chat window you already use, answering questions about your own studio as if it had read every number in the building. > Enterprises build a lake because they can hire the team to run it. You can't, and you shouldn't have to. So we hired it for you. ## What the model actually does This is why I said the model is the least interesting part. The AI (Claude, ChatGPT, whichever you use) is good at language. It's good at understanding what you asked and saying the answer back plainly. What it is not good at is finding the answer in a mess of disconnected systems that were never built to talk to each other. No model can, because the connections aren't there to find. Our job was to do that finding ahead of time. By the time you ask your question, the heavy lifting is already done: the data is gathered, the links are made, the picture is whole. The AI just reads it back to you. We supply the grounded, connected numbers. The model supplies the plain English. Neither half works without the other, and the half that took years to get right is the one you'll never have to think about. That's the whole design, and it's a deliberate one. Studios are not short on dashboards. They're short on answers. We didn't think the fix was another tool to learn, we thought it was doing the unglamorous work underneath so the answer could just appear. So that's what we built. --- # The business is the barrier, not the yoga or Pilates URL: https://kula.digital/blog/the-business-is-the-barrier-not-the-yoga Published: 2026-06-27 TL;DR: After four hours with teacher trainees, the pattern was clear: nobody fears the teaching, they fear the business around it — the lease, the pricing, the roster, the cash going out the door. That operating side, not the craft, is why most new studios fail, often taking friends-and-family money with them. What's changing is that a new owner can now start with an operating system watching the numbers that decide survival — the discipline of a 25-year operator, available on day one. It doesn't just help the people who start; it changes who gets to start at all. The business is the barrier, not the yoga or Pilates or the gym. More than nine in ten boutique fitness studios are not sustainably profitable. Only about half make money at all. I spent four hours today with a room full of teacher trainees, walking through the business of running a studio. Not one of them was worried about the teaching. They can teach. They've trained for it, they love it, they're good at it. What scared them was everything around it. The lease. The pricing. The roster. The money going out the door faster than it comes in. The part nobody trains you for. That's the real deterrent to starting a fitness business. It was never the craft. It's the operating side, and most people who can teach have no reason to be good at it. Now look at what it costs to find that out the hard way. Open a studio properly in a city like ours and you're into it for roughly half a million dollars. Most people assume that's equipment. It isn't. The build-out and the reformers are the smaller half. The bigger half is the working capital, the months of rent, payroll and overhead you carry before the membership base is stable enough to cover them. The money doesn't go on the thing you can see. It bleeds out quietly while you wait for the business to find its feet. And that money is rarely a bank's. It's rarely an investor's. It's the founder's own savings, plus cheques from the people who love them. Not arm's length capital with a term sheet and a board pack. Relational money. Which means the oversight is thin. There's an accountant doing the tax return months later, but nobody watching the numbers that decide survival while there's still time to act. So picture the actual situation. A first-time owner, brilliant at the craft, has put half a million dollars of mostly personal and family money into a business that bleeds cash for months before it stabilises, with inadequate oversight and no early warning system. The books get reconciled long after the decisions that mattered were already made. The ABS says 48% of new Australian businesses fail within four years. For sole traders, which is most studio founders, it's worse. When it goes, it doesn't just take a business. It takes the savings, the trust, and the next person who might have backed them. Scale that up and it stops being a studio story. Around 200 new boutique studios open in Australia each year, roughly A$100 million of capital deployed, and on the industry's own failure rates about half is ultimately destroyed. Call it A$50 million a year, in studios alone. Zoom out to small business across the country and you're looking at north of A$10 billion of invested capital destroyed every year. Same cause every time. Great operators who were never taught to run the business. For 25 years that was simply how it worked. You learned the business by losing money until you either figured it out or closed. The lucky ones had a mentor. Everyone else paid tuition in losses. What changed is recent. The shift in AI over the last six months made this possible. A new owner can now start with an operating system watching the numbers that decide whether they survive. Not a dashboard they have to learn to read. Something that tells them what matters this week, what's drifting, where the money actually is, and what to do about it. A clean view of the numbers that matter, while there's still time to act. The discipline a 25-year operator carries in their head, available on day one to someone who has never run anything. One trainee said something I haven't stopped thinking about. She said she'd start a business if that kind of support existed. That's not a customer asking for a discount. That's a person who counted herself out, counting herself back in. There's a whole population of people who would build something good if the business side didn't terrify them and didn't cost them their family's savings to learn. The support doesn't just help the ones who start. It changes who gets to start at all. So here's where I've landed. The technology is available now to support inexperienced owners and defend deployed capital. The next generation of studios shouldn't have to lose half a million dollars of trust to learn what was always knowable. Every new fitness business should start AI-augmented. Not eventually. From the first day. We call these Gen 2 studios. Kula Intelligence is being used in studios right now . http://kula.digital --- # Someone just resigned. What do I do? We asked the AI — live. URL: https://kula.digital/blog/gen-2-business-ep-01-teacher-resigned Published: 2026-06-21 TL;DR: In a live first look at Kula Intelligence, a boutique studio manager rated her teachers with the timeslot bias stripped out. Mid-session a teacher resigned — and within the hour the studio had a costed, zero-extra-spend replacement, a defensible retention decision, and a ready-to-run marketing campaign. No spreadsheets, just plain-English questions. This is episode one of an open build. Over the coming weeks I'm moving BodyMindLife — our boutique studio in Bondi — fully into AI, in public, and recording every step: the products, the decisions, and the part most people skip — getting the team on board. The whole thing becomes the launch playbook. Everything before Claude 4.8 was Gen 1 — a business run on instinct and spreadsheets. What comes after is Gen 2. BodyMindLife is on its way to becoming the first fully augmented AI boutique fitness studio, built on one belief: AI makes our studio more human, not less. > A note on privacy: this is a real session. All teacher names and a few identifying details have been changed so no one is identifiable. The numbers and the decisions are real. ## The problem every studio manager knows Becs runs the floor at BodyMindLife in Bondi. Her biggest recurring headache is the one almost no studio solves well: who is actually adding value to the schedule — and who isn't? Every owner thinks they know. The teacher with the big personality. The one who's been there longest. The Saturday regular. But "thinks they know" is not the same as knowing, and gut feel has a habit of protecting the wrong people and overlooking the quiet performers. So we did the obvious thing. We asked. ## The questions we put to Kula Intelligence You don't build a dashboard. You don't write a query. You ask, in plain English, the way you'd ask a sharp operations director who happens to have read every line of your data. The actual questions from the session: 1. "Assess this teacher as a yoga teacher. Rate all of our teachers and take into account peak and off-peak times. Define what the peak classes are at our Bondi studio — and do we have the best teachers on them?" 2. "This teacher just resigned. Who do we put on the Wednesday 6am class?" 3. "If we give her the double, what's your estimate of the improvement?" 4. "She's at risk of leaving for a competitor. Is it worth paying her an extra $10 per class?" 5. "Create the marketing — internal and social — to bring some wow to Wednesday mornings." Five questions. One conversation. No spreadsheets. ## What it found before anything dramatic happened First, it rewrote our definition of "peak." We'd always assumed weekday evenings were the prime real estate. The data said otherwise: demand at the studio is morning-led. Early weekday mornings and weekend mornings are where the rooms fill; weekday evenings are surprisingly soft. That single correction changes where your best teachers should stand. Second, it introduced a fairer way to rate teachers — measuring how many bodies each teacher draws above the average for that exact day and time. Put a teacher in a dead Tuesday-lunchtime class and a low headcount isn't their fault. Strip the timeslot effect out, and you finally see who genuinely pulls a crowd versus who's just been handed good classes. That's when the quiet performers showed up — strong teachers parked in off-peak classes, pulling well above their weight where almost no one was looking. And it's when the underperformers stopped being able to hide behind a flattering timeslot. ## And then a teacher resigned. Mid-session. We were looking at the studio's weakest draw — a teacher sitting on a prized Wednesday 6am class and filling barely a third of it. We were mid-sentence about what to do with him. Then Becs's phone buzzed. It was him. He was resigning. You could not script it. The exact problem we were diagnosing solved its first half on its own. The question instantly flipped from "how do we have an awkward conversation?" to "who do we put on Wednesday 6am — and how do we make it pull?" ## The solve: Plan A The Wednesday 6am class was never the problem — the teacher was. Same class, same time, with a strong teacher in the chair fills at more than double the rate. So you don't cut the class. You upgrade it. The replacement — call her Romy Winters — was already one of the studio's best-kept secrets. A high-drawing yoga teacher who'd never once been given a peak class. Moving her in costs the studio nothing extra: the wage for the class is fixed whoever teaches it. The numbers: - ~13 people across both Wednesday morning classes today, combined - 2× projected with a teacher like Romy - Cost per head cut from ~$17 to ~$9 — same wage bill Several thousand dollars a year of additional value, created out of thin air. That's the whole game: not spending more, but getting far more from money you're already spending. ## Make it a win-win-win A good decision helps one party. A great one helps everyone in the room. **For the studio:** a dead class becomes a peak class. Fuller rooms, better economics, and a second strong yoga teacher proven on mornings — exactly the bench depth every studio is short on. **For the teacher:** more classes, a small performance-linked rate bump (framed around the rooms she fills, never as a panic offer), public promotion of her classes, and a visible path to a bigger role. When a competitor down the road is circling, that is what keeps a good teacher — not a token raise, but feeling backed. **For the students:** their Wednesday mornings get genuinely better. A teacher who pulls a crowd creates the energy that makes people set the alarm. Full rooms keep members coming back. ## Whispers: the loop that makes it compound Here's the part we're most excited about. We're building BodyMindLife into a Kula-augmented studio — where the intelligence doesn't just sit with the owner, it reaches the teachers through Whispers: the small, member-level insights that help a teacher know their room. Use Whispers. Give us your honest feedback on them. And we'll get behind you — promote your classes, build your following, grow your role. Teachers who pay attention to their students get amplified. The studio gets sharper. Students get a more personal experience. Everyone's incentives point the same way. ## The marketing wrote itself Then we asked it to launch the new classes. Out came the campaign concept, ready to run: **WOW Wednesdays — "Winters on Wednesdays." Start your Wednesday wide awake.** A teaser, a reveal, a member email, in-studio posters, a launch-day social reel, and a "bring a mate free" offer that costs the studio nothing because the seats are currently empty — every guest is a free warm lead. A soft, quiet morning, turned into the one people circle on the calendar. ## What this actually means In under an hour, a studio manager went from "a teacher just quit" to a corrected, data-true definition of her best classes, an honest rating of every teacher with the timeslot bias stripped out, a costed zero-extra-spend replacement plan, a retention decision she could defend with numbers, and a finished marketing campaign. No spreadsheets. No dashboards. No analyst. Just questions, answered. > "It's not personal anymore. Instead of me saying 'I've noticed your classes are low,' the numbers are right there — I can just present it to them." > — Becs, Studio Manager That's the quiet revolution here. Kula Intelligence doesn't just tell you what's happening in your business. It hands you the conversation, the decision and the plan — and lets you get on with running the place. ## About this series I'm documenting the whole move of BodyMindLife into Gen 2 — the first fully augmented AI boutique fitness studio — in public, one episode at a time, so any owner can follow the same playbook. Coming up: getting the team on board, Whispers as a sixth sense for the room, rebuilding the week around real demand, and what Gen 2 actually feels like. We're looking for 20 early adopters to partner with — studios ready to run their next big decision on evidence, not instinct. See your studio today like you've never seen it before. --- # Becs didn't get a dashboard. She got a leadership team. URL: https://kula.digital/blog/becs-didnt-get-a-dashboard-she-got-a-leadership-team Published: 2026-06-21 TL;DR: A boutique studio manager spent half an hour with Kula Intelligence reading her studio's live data, and effectively had a CFO, an HR lead and a scheduling yield manager working beside her in real time. Together they turned a teacher's resignation into a costed plan that grew a quiet class, kept a valued teacher, and handled covers with care. Becs didn't get a dashboard last week. She got a leadership team. For about half an hour, the studio manager at a Bondi studio had a CFO, an HR lead and a scheduling yield manager working beside her in real time. The executive bench a boutique studio could never put on payroll, on tap, answering in plain English. And in those thirty minutes she turned what began as a stressful resignation into an exciting new chapter for a teacher and the studio. The team she sat down with was Kula Intelligence, the AI layer of KulaOS, reading the studio's own live data. No exports, no spreadsheets built the night before. Just real questions asked in plain language, and costed answers back. Here is what each of those three roles actually did. ### The yield manager found a star the roster had been hiding On headline fill, one teacher looked mid-pack at 46%. But two of her four classes sit in quiet Tuesday daytime timeslots that no one fills, so the headline was selling her short. Rate every teacher on a fair measure instead, bodies above or below the studio average for that exact day and hour, and she lands in the top six of 31. She had simply never been given a morning, which is where the demand at Bondi actually is. The roster had one of its best draws sitting in plain sight, waiting for the right room. That mattered, because a teacher had just let the studio know he was stepping back from two Wednesday morning classes. The instinct is to read that as a gap. The data read it as an opening. Wednesday 6am is a strong timeslot. A capable yoga teacher fills it well above 80%, and early weekday mornings average around 53% across the studio, while these two classes had been running closer to a quarter full. A quiet class ready to grow, and a hidden star ready for a morning. The match was sitting right there. ### The CFO priced the move in seconds Here is the part that makes the decision easy. The wage for the class is fixed either way, so moving the star into those Wednesday mornings costs nothing extra. It simply puts the spend behind a stronger draw. Modelled conservatively, that is roughly 600 additional visits across the year on a class that roughly doubles, at no added cost. The cost per head on the mat falls by close to half. Same money, a fuller room, a better morning. A real CFO would take a week and a meeting to land that. Becs had it before the kettle boiled. ### The HR lead handled the human moments with care Numbers are only half of running a studio. The other half is people, and this is where the day could have gone sideways without care. To back the teacher stepping into those mornings, the studio set a fair per-class rate that reflects the rooms she fills. Tied to delivery rather than to anyone's threat to leave, it sends the right signal to the whole team: we back the people who bring people back. A separate cover came up at the same time, a Pilates teacher heading off on holiday. The obvious replacement was one name. The data pointed to another, because her classes already drew the same members, so the handover felt familiar and the room stayed full. Members barely noticed the change, which is exactly the point. And a strong teacher planned three months overseas. Rather than brace for the loss, the studio asked who his students actually were. Of 168 over three months, three in four already trained with other teachers too, so they simply keep coming. What looked like a problem became a short, warm list and a clear cover. Check before you worry. ### Why it felt good, not cold The worry with this kind of tooling is that putting numbers on a people business makes it colder. The opposite happened. Becs put it best. "It's not personal anymore. Instead of me saying I've noticed your classes are low, the numbers are right there. I can just present it to them." The evidence does the difficult part, so the conversation gets calmer and fairer at the same time. It raised the humanity of the management rather than stripping it out. ### What actually changed Not a clever feature. Access. When an operator can ask their own business a direct question and get a direct, costed answer, the decisions that used to spread across a month of hedging fold into a single calm conversation. And they tend to be better decisions, because they are built on what really happened rather than on the loudest assumption in the room. "I never thought we'd get this much information," Becs said. For an operator, more information was never the goal. Better decisions, made faster, with less personal cost, is. There is a quieter prize underneath all of this, and it is the one owners feel. When the studio's judgment lives in the system and a capable manager can run it with this kind of backing, the owner can begin to step back from the day-to-day operational calls and spend their time working on the business instead of inside it. Not stepping away, stepping up. That is how an owner-run studio slowly becomes a business that holds its quality even when the owner is not in the room. Which, in the end, is the difference between owning a job and owning something worth far more. ---