Skip to content

Quick Start

Create an account using GitHub, Google, or email/password authentication.

Navigate to the dashboard and click New Agent. This opens the visual flow builder where you design your agent’s behavior.

Every agent needs at minimum:

  1. Start node — Entry point with a trigger type (manual for chat agents)
  2. Agent node — The LLM that processes input. Configure:
    • Model: Choose from 200+ models (e.g., openai/gpt-4o-mini)
    • System prompt: Instructions for the agent
    • Temperature: 0.3–0.5 for factual, 0.7–0.9 for creative
  3. End node — Terminates the flow and returns the response

Connect them with edges: Start → Agent → End.

Click Test in the flow builder to open the chat interface. Send a message and verify the agent responds correctly.

Activate your agent to make it available via:

  • Chat UI — Built-in web chat
  • API — Programmatic access with API keys
  • Platforms — Connect to Telegram, Discord, Slack, etc.
PatternUse CaseNodes
Simple ChatQ&A, supportStart → Agent → End
PipelineData processingStart → HTTP API → Agent → End
ConditionalMulti-purposeStart → Classify → If/Else → [Branches] → End
Multi-AgentComplex tasksStart → Agent Pool → End
ScheduledBackground jobsStart (cron) → HTTP API → Agent → End