Skip to content
ajan.io

Bring your agents inside your product.

With one API key, call models, run the agents your team built inside your own application, manage knowledge bases in code, and export usage data. Endpoints are provider-compatible, so the SDK you use today works once you change the base URL.

Start freeRead the docsSee the API surface

No credit card required.

Sample request · no real call is made

POST/agent/v1/chat/completions

curl -X POST https://api.ajan.io/agent/v1/chat/completions \
  -H "Authorization: Bearer $AJAN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "agentId": "agent_onboarding",
    "messages": [{
      "id": "msg_1",
      "role": "user",
      "parts": [{ "type": "text", "text": "Does annual leave carry over?" }]
    }],
    "stream": true
  }'

Response

Press Run to see a sample response.

One key, the whole platform

Everything you do in chat, agents, and workflows has an API counterpart.

Endpoints are grouped by function, and each is protected by its own scope. You open only the APIs a key needs: an integration key cannot read usage data, and a reporting key cannot publish an agent.

Completion APICalls OpenAI, Anthropic, Google, and Mistral models through provider-compatible endpoints./openai/{region}/v1/chat/completions
Embedding APIGenerates embeddings for semantic search and similarity./openai/{region}/v1/embeddings
Agents APICreates, updates, and publishes agents, and chats with them./agent/v1/chat/completions
Knowledge APIUploads files to knowledge bases, searches them, and shares access./knowledge/search
Integrations APIManages custom integrations, actions, and triggers./integrations/v1/create
Prompts and Skills APIManages the prompt library and skills in code./prompts/v1
Usage Export APIExports usage by user, agent, model, and workflow./export/users
Workflow Run ExportExports workflow runs and node results./workflows/{workflowId}/runs
Audit Logs APILists who did what, and when./audit-logs/{workspace_id}
User Management APIInvites users and updates roles./user-management/v1/invite

The SDK you use keeps working

Don’t rewrite your code — change the base URL.

Completion endpoints follow each provider’s official specification. Point your OpenAI, Anthropic, or Google client at ajan.io and requests go to the model in your workspace, with keys, model access, and usage tracking managed in one place.

  • OpenAI

    /openai/{region}/v1/chat/completions

  • OpenAI Responses

    /openai/{region}/v1/responses

  • Anthropic

    /anthropic/{region}/v1/messages

  • Google Gemini

    /google/{region}/v1beta/models/{model}:generateContent

  • Mistral

    /mistral/{region}/v1/fim/completions

Embed the agent your team built into your product

The team manages the agent; your application just calls it.

Share an agent with an API key and your application can talk to it. Instructions, knowledge, and actions stay in the agent; when the team updates it, your code does not change. The endpoint uses a Vercel AI SDK compatible message format and streams the answer back.

  1. The team builds the agent

    Instructions, knowledge, and actions live in the panel.

  2. It is shared with a key

    The agent is opened to the app’s API key.

  3. The app calls it

    One request with the agentId and the message.

  4. The user sees the answer

    Streamed back, together with its source.

See how agents are built

Manage knowledge bases in code

When documents change in your system, the knowledge base stays current.

With the Knowledge API you upload files, update attachments, search within a knowledge base, and share access. Carry changes from your document management system into ajan.io with a workflow, and use search results in your own application.

Scoped keys, clear limits

Every key carries just enough permission for its job.

Personal keys only use the completion APIs. Workspace keys are created by admins, each with its own scopes. Rate limits apply per workspace, model, and key; keys that use the same model share its tokens-per-minute quota.

  • Personal key

    A developer’s own experiments; completions only.

  • Workspace key

    For applications; an admin sets its scopes.

  • Access by sharing

    A key only calls agents shared with it.

  • Limits and monitoring

    Usage is visible per key and can be exported.

Your usage and audit data is yours

Let finance, compliance, and analytics teams see the data in their own tools.

The Usage Export API exports usage by user, agent, model, project, and workflow; workspaces on their own keys also get cost and token detail. The Audit Logs API returns admin actions, and the Workflow Run Export API returns node results for every run.

  • Finance

    Allocate cost by department and project.

  • Compliance

    Keep admin actions in your own archive.

  • Analytics

    Track agent adoption in your BI tool.

The same API in the cloud and on a dedicated deployment

The environment changes; the contract does not.

On the multi-tenant cloud, the base URL is api.ajan.io. On a dedicated deployment the same endpoints run under your own domain; you change only the base URL in your code. Recipes and sample projects are in the docs.

Read the API docs

  • ajan.io cloud

    https://api.ajan.io

    Multi-tenant, default

  • Dedicated deployment

    https://<your-domain>/api/public

    For enterprise deployments

Questions about the API

What developers ask before they start building.

All FAQs

Plans on the pricing page cover users, agents, and AI Credits. API calls draw usage based on the models you enable; if you bring your own key, the provider bills you directly.

Send your first request today.

Create a key in a free workspace; talk to our team if you need a dedicated deployment or higher limits.

Start freeTalk to sales