Docs/Build/Agent Flow

Agent Flow

Updated Mar 29, 2026Build

Build AI-powered automation workflows with agents, triggers, and actions.

Overview

Agent Flow lets you build AI-powered automation workflows. Define agents that execute tasks, connect them with triggers, and chain actions together.

Key Concepts

  • Agent — An AI-powered unit that performs a specific task (e.g., summarize news, send a message, analyze data)
  • Trigger — What starts a workflow (e.g., schedule, webhook, manual run)
  • Action — What the agent does (e.g., call an API, send a Telegram message, write to a database)
  • Workflow — A chain of agents connected by data flow

Workspace Tabs

  • Overview — Workflow diagram and agent configuration
  • Runs — History of all workflow executions with status and logs
  • Inbox — Messages and outputs from your agents
  • Logs — Detailed execution logs for debugging
  • Code — View and edit the generated agent code
  • API — Auto-generated API documentation for your workflow
  • Memory — Persistent memory that agents can read and write

Built-in Tools

Agents have access to built-in tools:

  • HTTP requests — Call any API or webhook
  • Send messages — Telegram, Slack, Discord, Email
  • Web search — Search the internet for information
  • File operations — Read and write files
  • Database — Query and update data

Example Use Cases

  • Daily news digest sent to Telegram
  • Monitor a website and alert on changes
  • Analyze customer feedback and create reports
  • Automate data collection from multiple APIs
  • Scheduled social media posting