Compose your own

Builder

Toggle modules, pick a design, preview live, submit for delivery.

CRMon
Bookingon
Invoicingon
Reviewsoff
Loyaltyoff
Open builder
A la carte

Modules

18 individual operational tools at $29-$49 each. Mix and match.

See all
Visual identity

Templates

11 aesthetics. Same kit, different look. Compose orthogonally.

See all
Free forever · MIT

The three free cores

Each is a Vercel-deployable Next.js project.Browse all cores →
Home/Code-to-Own AI
The AI Layer

AI you own, not AI you rent.

Every other AI product locks you into their hosted model and routes your prompts through their servers. LaunchKits ships AI the opposite way: as opt-in modules you fork and own. You bring your own model and API keys, the code lives in your repo, and we never proxy your AI requests and never see your prompts. Today that means the $9/mo AI Assistant. The AI-native modules are on the roadmap, and honestly labeled as such below.

Bring your own model

Model-agnostic through one gateway key: OpenAI, Anthropic, Google, xAI, DeepSeek, or Mistral. You choose; you can switch any time.

Bring your own keys

The module reads your own API key from your own deployment. No shared LaunchKits key, no vendor lock-in on the credential.

We never see your prompts

The request goes straight from your server to your provider. LaunchKits runs no proxy and no AI server, so it is never in the path.

You own the code forever

The AI module source ships to your private GitHub repo, code-to-own like every other module. Cancel anything and it keeps running.

Owned AI vs the hosted AI you rent

The difference is who holds the keys, the code, and the request path. Every LaunchKits row below is a property of the shipping AI Assistant module, verified against its source.

DimensionHosted AI-SaaSLaunchKits (code-to-own)
Whose API keyTheirs, baked into their platformYours: your own gateway key in your own deployment
Who picks the modelThey do (one vendor, locked)You do: OpenAI, Anthropic, Google, xAI, DeepSeek, or Mistral
Who sees your promptsTheir servers route every requestNobody at LaunchKits; the call goes from your server to your provider
Who owns the codeThey do, and you rent accessYou do: the module source ships to your GitHub repo
What happens if you cancelThe AI stops workingYou keep the code and keep running it on your keys
What the monthly fee buysAccess to their hosted AIMetering headroom only, never a gate on code you bought
Available today

AI Assistant

The $9/mo AI add-on. Ask natural-language questions about your own business data, get AI-generated weekly and monthly reports, and fire smart alerts on overdue AR, new customers, and payments. Model-agnostic through one AI Gateway key, metered per business against the Core token bucket. It grounds every answer by calling your installed modules, never by touching their raw tables.

$9/mo
Add-on metering only. You pay your own provider for tokens.
What it does
  • Natural-language queries over your own business data
  • AI-generated weekly and monthly reports
  • Smart alerts on overdue AR, new customers, and payment events
  • Model-agnostic via one AI Gateway key (OpenAI, Anthropic, Google, xAI, DeepSeek, Mistral)
  • Metered per business against the Core token bucket, gated before every call
  • Grounds answers through installed modules’ lookups, never their raw tables
  • Keyless demo mode runs a simulated stream when no gateway key is set
On the roadmap

These are the AI-native cornerstones we are building toward. They are not available yet, and we will not sell them as if they were. When each ships, it ships the same way: your keys, your model, your code.

AI ChatPlanned

An embeddable, model-agnostic chat surface over your own data and docs.

Brain OSPlanned

A durable knowledge and memory layer your modules can read and write.

Agent RuntimePlanned

Scheduled and triggered agents that act across your installed modules.

The rule that makes it real

Your prompts never touch our servers.

Because the AI module runs inside the code you own, the request path is yours end to end: your deployment, your gateway key, your provider. LaunchKits runs no proxy and no AI endpoint, so there is nowhere for us to read a prompt even if we wanted to. And with no key configured, the assistant runs a local demo and calls nothing external at all. That is the whole point of owning the AI instead of renting it.

AI ownership questions

Do you ever see my prompts or my business data?+

No. The AI module runs inside your own deployment and calls your chosen provider with your own API key. LaunchKits operates no proxy and no AI server, so your prompts and data never pass through anything we run. When no key is set, the module runs a local demo stream and makes no external call at all.

Which AI models can I use?+

You bring your own. The module is model-agnostic through a single AI Gateway key that resolves a "provider/model" string, so one key reaches OpenAI, Anthropic, Google, xAI, DeepSeek, or Mistral. You pick the model in settings; swap it any time. You supply the credentials, so you control the vendor relationship and the bill.

Do I own the AI code, or is it a license I can lose?+

You own it. The AI module is delivered as source into your own private GitHub repository, the same code-to-own delivery as every other module. Once it is in your repo it is yours to run, fork, and keep. Nothing we change on our side can take it back.

What does the $9/mo AI Assistant fee actually pay for?+

It raises the monthly token headroom metered against your business, nothing more. It is not the cost of the AI itself, since you pay your own provider directly for tokens. And it never gates a purchased feature: the module code keeps working whether or not the add-on is active.

Are you approved or endorsed by OpenAI, Anthropic, or the other providers?+

No, and we do not claim to be. The module simply lets you bring your own account with any of those providers. There is no partnership or endorsement implied. You own the relationship with whichever vendor you choose.

Browse the Marketplace

Compare the whole model: own vs rent · FAQ