The AI brain, your way

Use your own AI.
$0 markup.

Already using an AI provider you like — or running your own? Point your assistant straight at it. We never touch your data and never add a markup, and the included AI stays the easy default if you'd rather not.

$0.00
Markup on your tokens
Any
Provider or endpoint
Your
Rate limits, your quota
Anytime
Switch anytime, instantly
How it works

Three steps to your own AI

Connect it once, set your limits, and we handle the rest. Your AI, your bill, your rules — we never run your prompts through a shared account.

01

Connect your AI provider

Paste an API key from any provider, or point us at a base URL — a hosted gateway or your own vLLM / Ollama box. We store it encrypted and scoped to your account.

02

Set it as the default and cap the spend

Make it your assistant's default and set hard limits on spend and how many run at once. Mix the included AI and your own across your team of assistants — per assistant, or per kind of job.

03

It runs on your account, you stay in control

Every request goes straight to your provider on your own account. We never train on, store, or run your prompts through a shared account. Your limits, your bill.

Supported providers

Works with the AI you already use

Hosted gateways, the major clouds, fast specialty providers, or a server you run yourself. If it offers an AI endpoint, your assistant can run on it.

AnthropicAPI key
OpenAIAPI key
AWS BedrockIAM role
Google VertexService account
TogetherAPI key
GroqAPI key
vLLMSelf-hosted
OllamaSelf-hosted
Two ways to run

Managed by default. Bring your own when it pays.

Most teams use the included AI and never look back. Bring your own once you have a provider deal or your own setup to put to work. You can mix both across your team of assistants.

Included AIRecommended

We run the AI for you — no keys to manage, no capacity to plan, nothing to keep online.

  • Ready instantly — no setup, no keys to manage
  • Top-tier and fast models, scaled for you automatically
  • One bill — you only pay for what it uses
  • We handle rate limits, failover & capacity
Bring your own modelFor power users

You keep your own account with the provider. We just connect to it — honestly, and at cost.

  • $0.00 markup — you pay your provider directly
  • Your own rate limits and provider discounts
  • Any provider, region, or self-hosted endpoint
  • You manage keys, quota, and capacity
AI options you can pick for each assistant
Managed · FrontierManaged

Claude Opus & Sonnet — top-tier AI, included and ready. No setup; you only pay for what it uses, billed with your plan.

Managed · FastManaged

A faster, lighter AI for high-volume work at a fraction of the cost.

BYO · Anthropic keyBYO

Advanced: bring your own Anthropic key — we never mark it up, and you keep your own limits.

BYO · Any endpointBYO

Advanced: works with your own AI provider — OpenAI, Together, Bedrock, or a self-hosted server.

Run your AI your way

Start on the included AI in seconds, or connect your own keys when you're ready. Either way, we handle the hard parts.