AI & bring your own key
Use the AI account you already have. No markup.
Every workspace connects its own AI provider key, and your provider bills you directly at their price. We sell no AI credits and add nothing on top. Your plan pays for the platform, not for tokens.
What you get
AI costs you can see and control.
Pay your provider at cost
Token usage goes to the account that owns the key, at the provider's published prices. There are no AI credits to buy or run out of.
Choose the model per step
Each AI step picks its own model from the providers you've connected. Browse a provider's live model list or type any model ID it supports.
Free testing
Dry Run simulates AI steps with placeholder output, so testing a workflow never calls your provider.
Cost per step
Run metrics show tokens in and out and an estimated AI cost for each node, so you can see which step is expensive.
A built-in cost cap
Every server run stops if its estimated AI cost passes $2, to protect you from a runaway loop.
Keys stay protected
Keys are checked with the provider before saving, encrypted at rest, and never sent back to a browser. Settings shows only a short hint.
Supported providers
14 providers. Connect one or several.
Connect a key in Settings → AI Providers. It's checked with the provider before it's saved, then stored encrypted.
- OpenAI
- Anthropic
- Google Gemini
- Mistral
- Groq
- xAI
- DeepSeek
- OpenRouter
- Together AI
- Fireworks AI
- Azure OpenAI
- AWS Bedrock
- Google Vertex AI
- Any OpenAI-compatible endpoint
| Feature | Needs your AI key? |
|---|---|
| Agent, Observer and AI Generate steps in a real run | Yes |
| Marketplace agents | Yes |
| Canvas insights and AI tips | Yes |
| Dry Run | No. AI steps are simulated |
| Triggers, integrations, routers, validators, outputs, human review | No |
| Run history, analytics and docs | No |
Good to know
- Custom endpoints and Azure endpoints must be public https addresses. Local or private network addresses are blocked for security.
- Cost estimates use list prices for popular models and are rough for others. Your provider's bill is the source of truth.
FAQ
Questions, answered.
What will AI cost me?
Whatever your provider charges for the tokens you use, billed by them directly. We add no markup. Each run shows an estimated AI cost, and you can set a spending limit on the key with your provider for a hard ceiling.
Can I connect more than one provider?
Yes. Connect as many as you like, then pick the model for each AI step. A real run is blocked, with a clear message, if a step uses a provider you haven't connected.
Do I need an AI key for everything?
No. Triggers, integrations, routers, transformers, validators, outputs and human review work without a key. Agent, Observer and AI Generate steps, marketplace agents and AI tips need one.
Who sees my prompts?
Prompts go to the AI provider you connected, under your own account. Your agreement with that provider, including its data-retention and training settings, applies.
Can I use a self-hosted model?
Yes, through the custom OpenAI-compatible endpoint, as long as it's reachable at a public, secured https address.
Connect your key and run your first AI step.
Build and Dry Run on the Free plan, no credit card. Run live on a paid plan, on your own AI key at cost.
Free to explore · No credit card · 30-day money-back guarantee on paid plans