Skip to main content

Deploying Agents

Once your agent is tested and ready, deploy it for production use.

Deployment Options​

REST API Endpoint​

Deploy your agent as a REST API that any application can call:

  1. Open your agent in AI Studio
  2. Click Deploy
  3. Copy the API endpoint URL and API key

Send a message:

curl -X POST https://app.onflowstack.com/api/v1/agents/{agent-id}/chat \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "message": "What is the status of order ORD-12345?",
    "session_id": "user-abc-123"
  }'

Response:

{
  "response": "I found your order ORD-12345. It was shipped on April 18 via FedEx (tracking: 7891234567). Expected delivery is April 22. Is there anything else I can help with?",
  "tools_used": ["lookup_order"],
  "tokens_used": 342,
  "session_id": "user-abc-123"
}

Embedded in Workflows​

Use your agent as a step in an automation workflow:

  1. Open a workflow in the Workflow Builder
  2. Add an AI Agent node
  3. Select your deployed agent
  4. Connect inputs (user message) and outputs (agent response)

Example: Auto-respond to support emails

Gmail Trigger (New Email) → AI Agent (Support Bot) → Gmail (Send Reply)

Chat Widget (Coming Soon)​

Embed a chat widget directly on your website that connects to your FlowStack agent.

Monitoring Deployed Agents​

Track agent performance in the dashboard:

  • Total conversations — Number of unique sessions
  • Messages processed — Total messages sent and received
  • Average response time — How long the agent takes to respond
  • Tool usage — Which tools are used most frequently
  • Token consumption — API costs per agent
  • Error rate — Failed responses and tool execution errors

Scaling​

Agents scale automatically based on demand:

  • Free plan — 100 agent messages/month
  • Pro plan — 5,000 agent messages/month
  • Team plan — 25,000 agent messages/month
  • Enterprise — Custom limits with dedicated infrastructure

Security​

  • All agent communications are encrypted in transit (TLS 1.3)
  • API keys are required for all agent endpoints
  • Tool executions are sandboxed — agents cannot access systems beyond their configured tools
  • All interactions are logged for audit purposes
  • Sensitive data can be redacted from logs on Enterprise plans