We built an AI that actually knows your studio.

Every studio owner we've worked with has the same shape of problem. The data exists — it lives in Mindbody or Hapana, in your Xero ledger, in the ClassPass dashboard, in a CSV someone exported last Tuesday. It just doesn't talk to itself. Revenue is in one tab, member health is in another, class attendance is in a third. And the AI tools that could stitch this together are brilliant at general advice and useless at specific answers, because they cannot see the numbers.

Kula Intelligence is the fix. It plugs the tools you already use into the AI client you already use — Claude Desktop, ChatGPT, Cursor, Lovable — and gives that AI a safe, scoped, audited window into your studio's data. Plus enough business context to give answers that sound like you wrote them.

How it actually works

Sign up takes about ten minutes. A short wizard walks you through connecting the systems you already run on. Live today: Mindbody, GymMaster, Wix, ClassPass, Stripe, Xero, QuickBooks, Meta, CSV imports — with Hapana, Mariana Tek, Square, Squarespace, MYOB and a handful of marketing sources in the queue behind them. Connect one or connect six; the more Kula can see, the more of your studio it can join up.

Then you close the laptop. Kula reads everything you connected — the full history, not just this month — and emails you when it's done. That email has a connection link already made for you: one click, confirm it in Claude, and Kula is sitting in the chat window you already use. Nothing to mint, nothing to paste.

On ChatGPT it's a slightly longer road, and we'd rather say so than surprise you. ChatGPT has no one-click install for connectors like Kula yet, so it's a short manual add — turn on Developer mode, hit Create, paste the link from the same email. Two minutes. We're in OpenAI's app directory queue; when that lands, it becomes one click there too.

Then you say “let's get going”

This is the part we spent the longest getting right. A connected AI with access to your data is not the same thing as an AI that understands your studio, and the gap between those two is where most tools quietly die.

So the first thing Kula does is sit down with you. You type let's get going and it starts reading your studio back to you — here's what I think your peak hours are, here's what I think a regular looks like here, here's where your numbers go quiet and I don't know why. You correct it. It remembers. That first step takes about ten minutes; the whole opening arc is around forty, and you can stop anywhere and pick it up next week without repeating yourself.

By the end you have a playbook built from your own numbers rather than someone's idea of a boutique studio — and Kula has enough context that questions like these get real answers:

• Who'd love a check-in from me this week?
• How is the reformer cohort tracking — are they staying with us?
• Which classes are quietly growing, and which could use a refresh?
• Draft a warm note to the twelve members I haven't seen since May.

See what the guided sessions cover →

You don't learn a new dashboard. You don't babysit another tool. The interface is the chat window you already use — only now it knows your numbers.

Studios are not short on dashboards. They're short on answers. We replaced the dashboard with a conversation.

Per-studio, by construction

The architecture under the hood is the part we're quietly proud of. Every studio gets its own database. There's no “cross-studio” read path. The AI cannot accidentally answer your question with another studio's data, because that data is not even in the same database. Every read and every write goes through a purpose-built tool. Every action is audited. You can revoke any AI client's access in one click.

That database sits in the region you operate in — Australian studios on Australian infrastructure, US studios on US infrastructure — and it's the same boundary whether it's you asking, your studio manager, or a coach you've given a narrower view to. Each person connects with their own link. You can see who has access, and cut any of it off, from the console.

Kula will also tell you what it doesn't know. Ask it something your connected systems genuinely can't answer — GymMaster has no financial API, for instance — and it says so, rather than doing the arithmetic on the fragment that did arrive and handing you a confident wrong number.

We think trust is the actual feature.