Skip to content

Quickstart

This guide walks you through signing up, creating your first agent, and seeing it respond — in under five minutes.


  • An AgenFleet account (sign up at agenfleet.ai)
  • An Anthropic API key (or other supported provider) — agents use your own key; token costs go directly to your provider account
  • A Telegram account (for receiving agent output — optional but recommended)

  1. Log in and connect your API key

    Go to portal.agenfleet.ai and sign in. First-time users land on the onboarding wizard — it takes about 90 seconds to complete. During onboarding you’ll add your Anthropic API key under Settings → Integrations. This is how your agents access AI models — token costs go directly to your Anthropic account.

  2. Create your first agent

    From the dashboard, click New Agent. Give it a name (e.g., Research Assistant) and select a role template. Templates pre-configure the model, tools, and a starter SOUL file so you don’t need to start from scratch.

    Available starter templates:

    • Research Assistant — web search, summarization, report drafting
    • Ops Monitor — system health checks, alerting, log analysis
    • Content Writer — drafting, editing, scheduling social posts
    • Data Analyst — data summarization, trend detection, charting
  3. Review the configuration

    The wizard shows you the generated config. Key fields to check:

    • Model — defaults to Claude Haiku 4.5 (fast, cost-efficient). Upgrade to Sonnet or Opus for complex tasks.
    • Instructions — a short paragraph telling the agent what it does. Edit this to match your use case.
    • Tools — which capabilities the agent has access to (web search, file access, etc.)

    Click Save Agent when ready.

  4. Send your first message

    In the agent’s chat panel, type a prompt and hit send. For example:

    Summarize the top 3 AI industry news stories from today.

    The agent will respond within a few seconds. You’re talking to a live, deployed AI agent.

  5. Connect a notification channel (optional but recommended)

    Go to Settings → Notifications and link your Telegram account or Slack workspace. Once connected, any scheduled task output from this agent will be delivered directly to you.


When you clicked Save Agent, AgenFleet:

  • Provisioned a containerized agent process on isolated infrastructure
  • Loaded your SOUL file (the agent’s identity and instructions) into its context
  • Set up a private memory store for the agent
  • Exposed it for real-time chat and scheduled cron jobs

Your agent is now live and ready to be put on a schedule.