Developers

Talk to your agents from your own code

Every agent you build is callable. Send a message, get a structured response, and keep the same instructions and access rules.

  • HTTPS + bearer token
  • Same guardrails as the app
  • Pro and Enterprise
1 call

To send an agent a message

Same

Instructions and integration scope

Counted

Against your plan allowance

Simple request shape

Authenticate, name the agent, send a message. The response includes the agent's reply and the run identifier that shows up in your activity log.

  • HTTPS + bearer token auth
  • One endpoint per agent message
  • Runs counted against your plan allowance

Same guardrails

API calls respect the agent's instructions and attached integrations — there is no separate, less-restricted path.

Call your agents from your own product

Create a workspace, build the agent in the UI, then wire it into your code.

Start free

Getting access

API access is available on Pro and Enterprise. Create your workspace, then request a key from your settings.

Common questions

Is there a less-restricted API path?

No. API calls run the same instructions, the same attached integrations and the same approval rules as the app, and land in the same run log.

How do I get a key?

Create your workspace on Pro, then request a key from your settings. Keys are scoped to your workspace and can be rotated.

Do API runs count towards my allowance?

Yes, one message is one request — the same as chat, schedules and webhooks.

Describe your first agent today

Free to start. No credit card, no setup calls, no engineering ticket.

Get started free