Documentation
Tutorial

Quick Start Guide

Understand Supervity’s core concepts and run your first agent

This guide introduces how Supervity works, not just how to click through the UI.

By the end of this page, you will understand:

  • what an agent (operator) is
  • how jobs are executed
  • how auto apps, integrations, approvals, and schedules fit together

All roles share this foundation.

Core Concepts (Read This First)

Agent (Operator)

An agent is an AI operator that:

  • understands your intent
  • plans how work should be done
  • executes jobs using tools and policies

Agents don’t blindly automate. They operate within human-defined boundaries.

Job

A job is a single execution of work:

  • triggered manually, on a schedule, or via API
  • composed of multiple steps
  • observable, retryable, and auditable

Every agent action ultimately results in jobs.

Auto App

An auto app is a packaged operational capability:

  • workflows
  • tools
  • policies
  • approvals
  • integrations

Agents use auto apps to perform real work.

Governance Primitives

These apply to all roles:

  • integrations (what tools agents can use)
  • approvals (where humans intervene)
  • schedules (when work runs)

Step 1: Sign Up & Workspace Setup

  1. Log in to Supervity
  2. Set up your workspace
  3. Define who can create agents and approve actions

This establishes your execution boundary.

Step 2: Connect an Integration

Agents need tools.

  1. Go to Settings → Integrations
  2. Connect a service (email, crm, chat, storage)
  3. Review and approve permissions

Integrations become tools agents can use inside jobs.

Step 3: Create Your First Agent

  1. Click Create Agent
  2. Describe the outcome you want

Example:

"Send me a daily email summary of new leads"

The agent will:

  • plan the job
  • select tools
  • define execution steps
  • surface approval points

Nothing runs without your approval.

Step 4: Test Execution

Run a test job to:

  • observe step execution
  • inspect inputs and outputs
  • verify approvals and retries

This is where trust is built.

Step 5: Operationalize

Choose how jobs should run:

  • manual
  • scheduled
  • API-triggered
  • event-driven

Execution always respects governance.

Where to Go Next

start-here/02-quick-start.md