Skip to content
Vonzio docs

Getting started

This page assumes you have an account on vonz.io. If you don’t, register first.

The model graph

Section titled “The model graph”

Before clicking around, internalize this:

You ─ have many ─→ Agent profiles
                   │
                   │ each one runs in
                   ▼
                   Workspaces ─ live in ─→ Docker containers
                   │
                   │ talk to
                   ▼
                   MCP servers   (memory, notify, gmail, teller, vonzio)
                   Tools          (Bash, Read, Write, custom)
                   Integrations   (Slack, Telegram, Gmail, Teller, Email, Webhook)

Read Architecture for the long version. The list below is the minimum to go from zero to a working agent.

  1. Add a model API key.

    Settings → Keys → add your Anthropic API key (or Ollama URL if you self-host a local model). The platform routes agent traffic through this key — your key, your spend.

  2. Create an agent profile.

    Settings → Profiles → New profile. Pick a model, give it a slug (coding, vzfinance, etc.), and accept the defaults for now. The slug is how you’ll address it in chat (@coding) and in API calls.

  3. Open the dashboard.

    The chat surface is the default landing page. Pick your profile from the dropdown, type a message, hit send. The agent runs in a fresh Docker container, streams tokens back over WebSocket, and ends the turn when it has nothing more to say.

  4. (Optional) Connect an integration.

    Settings → Integrations. Slack and Telegram give you a chat surface outside the dashboard. Gmail and Teller give the agent read access to data sources. See Integrations overview.

  5. (Optional) Schedule a playbook.

    Profiles → your profile → Playbooks. A playbook is a scheduled prompt that runs autonomously on cron. Pair with notifications and you’ve got a proactive agent. See Playbooks.

Section titled “What to read next”
  • Quickstart — concrete walkthrough that ends with a working agent doing something useful
  • Architecture — the mental model. Don’t skip if you plan to use the API.
  • API authentication — how to mint a token for CLI / scripts / agents