Skip to main content
Deck v2 is in preview with closed access. API and console URLs are not yet publicly available.
Deck agents log in, navigate, and operate software exactly like a human does. No APIs on the other end, no custom integrations, no limits. The Deck API lets you programmatically spin up Computer Use agents, execute tasks against any website, and return structured results to your application.

​
Why use Deck

  • Works on any software. If a human can use it, Deck can automate it. Create a source from any URL and start running tasks on day one.
  • Structured, deterministic output. Every task returns schema-validated JSON. No parsing HTML, no fragile scraping, no unpredictable AI output.
  • One agent or thousands. Deck provisions isolated browser sessions on demand. Go from one agent to hundreds running in parallel with no config changes.
  • Full audit trail. Every click, every decision, every recovery is logged. Credential vault, human-in-the-loop interactions, and the ability to cancel any agent instantly.
  • Event-driven by default. Results are pushed to webhooks, SQS, Pub/Sub, or any supported destination.

​
Get started

Quickstart

Create an agent, connect a user, and run your first task.

Concepts

Agents, sources, connections, tasks, and events.

API reference

Endpoints, request formats, and response examples.

​
Explore

How Deck works

Platform architecture, browser orchestration, and the execution model.

Using the API

Authentication, pagination, idempotency, and error handling.

Build with AI

Ready-made context for building with AI coding agents.

Events and destinations

Subscribe to events, configure destinations, and track deliveries.