Bring your own AI key.

When your app needs AI inside it — to summarize, to write, to classify — you plug in a key from a provider you trust (OpenAI, Anthropic, Google, others). Hatchable doesn't resell AI by the token. You pay providers directly, at their published rates.

What's an API key, really?

It's a long random string that proves to a company like OpenAI that requests are coming from your account, so they can charge you for what gets used. It looks something like:

sk-proj-aBCdef1GhI2jKlMnoPQRstuv…

You go to your AI provider's website, click "create a key", they show it to you once, and you paste it into Hatchable. From that moment on, when your app does something AI-powered, the bill goes to your provider account — not to us.

If you've ever used Mailchimp with your own SMTP credentials, or pointed a tool at your own Twilio account — same idea.

Why we don't bill you for AI ourselves

It costs more, you have less control, and we'd be a middleman taking a cut on every word the AI generates.

By having you bring your own key, three things become true:

You pay providers their published rates. No markup from us. AI pricing changes constantly, and reselling means we'd have to bake in margin, which means you'd pay more for the same thing.
You can use whatever provider you want. OpenAI today, Anthropic next month, a local model when those get good. Switch by swapping a key. The app you built keeps working.
You can see exactly what's being spent. Your provider's dashboard shows the truth — every token, every dollar. Spending caps live on their side, where they're enforceable.

The trade-off: you have to make a provider account once. After that it's invisible.

Where the key lives, and what we do with it

Encrypted on our servers. The raw key never leaves the database except through the encrypted SDK call your app makes to the provider. It's not readable in logs, not visible in the console after you save it.
Never sent to your app's code. When your app says "summarize this", the platform forwards the request to the provider with your key attached — your app code never touches the actual key string. This is on purpose. It means a bug in the app can't accidentally print or leak it.
Always revocable. Open the project's settings, hit "remove key", it's gone. Even faster: revoke it on the provider's site (OpenAI, Anthropic, etc.) and the key stops working everywhere immediately.
Never shared between projects. Each project has its own slot. Setting a key on your chore tracker doesn't expose it to your reading log.

Three ways an app can use AI

How keys are scoped depends on what kind of app it is. You'll see the difference when you set one up — the app tells you which one it expects.

Just for me

One key, project-wide

The simplest setup. Your app uses your key. Anyone you invite to use the app shares the bill (which goes to your provider, not Hatchable).

Right for: personal tools, family apps, internal team tools where you're happy to cover the AI spend for everyone you let in.

Example: a meeting-notes summarizer your team uses. You paste in your OpenAI key. The whole team can summarize. Your OpenAI bill reflects everyone's usage combined.

Across everything you build

One key, account-wide

Same key shared across all your projects. Add it once at the account level, and every app you build can use it without you re-entering it. Saves you a paste each time.

Right for: people building lots of small tools who don't want to manage keys per project.

Example: you've got a reading log, a workout tracker, and a habit checker. Each one might use AI to give you weekly insights. One account-level OpenAI key powers all three.

Each user brings their own

Key per user (multi-tenant)

For apps you want other people to copy and run as their own thing. Each user who makes a copy sets up their own AI key. They pay for their own usage. You're not on the hook for the AI bill of strangers.

Right for: building something close to a SaaS — a worksheet generator for teachers, a brand monitor for marketers, a journal helper for writers. Anyone can grab your template and run their own private copy.

Example: you publish a "make your own" worksheet generator template. A teacher in Ohio clicks "make my own copy," puts in their own OpenAI key, and uses the app forever — without ever sending you a dollar or showing up in your provider bill.

You don't have to pick the type — the app you're using or building tells you which one it needs. If you're starting from a template, the template's already configured.

Adding, switching, or removing a key

Adding

Open your project, hit Settings, find the "AI keys" section. Paste the key, hit save. You can paste keys from any of these providers:

OpenAI — the GPT models. Get a key at platform.openai.com/api-keys
Anthropic — the Claude models. Get a key at console.anthropic.com
Google — Gemini. Get a key at aistudio.google.com/app/apikey

Switching

Most apps don't care which provider answers — they ask for "the smartest available" or "the fastest available", and whichever provider you've set a key for handles the request. So switching from OpenAI to Anthropic is usually a paste-and-save. Apps that need a specific model (some video or voice tools) will say so.

Removing

Hit "remove key" in Settings. The slot is empty again. The next time the app tries to use AI, it'll show a "needs setup" screen until a key is back in place.

For maximum safety, also revoke the key on the provider's website. That ensures it can't be used anywhere — even by us, even by accident.

What it costs you

It depends on what your app does. A useful rough order of magnitude:

A page-of-text summary with a smart model: cents.
A few hundred summaries a day: a few dollars a month.
A live chat assistant with a thousand messages a day: tens of dollars a month, depending on length.

Provider websites have calculators. They also let you set monthly spending caps — we recommend starting with a small one ($5–$20) and bumping it up if you actually hit the ceiling. Caps live on the provider side because that's where the billing happens, and where they can actually be enforced.

Common questions

Do I need an AI key for every Hatchable app?: No. Most apps don't use AI inside themselves. The AI that builds your app for you is included free. You only need an AI key when the app you're building is going to call AI as part of doing its job — like a summarizer, a chatbot, or a content rewriter.
Will my key power someone else's app?: No. Keys are scoped to the project (or account, or user) where you set them. There's no cross-project sharing.
Can I see what's being spent?: On the provider's website, yes — they keep the source of truth. Hatchable doesn't track your provider spend (that's not our data to collect).
What if my key gets compromised?: Revoke it on the provider's site immediately — that kills it everywhere, including in Hatchable. Then generate a new one and paste it in. Treat keys like passwords; don't share them in screenshots or send them in chat.
Can I use a free-tier provider key?: Yes, if the provider issues them. Google AI Studio has a generous free tier — same paste flow. Free tiers usually have lower rate limits, so the app may feel slower under load.
What about cost surprises?: Set a monthly spend cap on your provider's site. They enforce it. If your app suddenly does something pathological, the provider stops accepting requests at the cap and you get an email. The worst case is "the app stops working until next month" — never a six-figure bill.

Bring a key, ship anything.

Once a key's pasted, every AI capability your app might need just works. You're never paying us for tokens.

Get started →

← Your AI agent Next: Sharing & selling →