Core concepts

Stunt Double is organised around a small number of building blocks. This page explains each one and how they fit together.

Workspaces and projects

A workspace is the top-level container for your team. It owns members, billing, and every project. A project sits inside a workspace and groups the actors, checklists, and runs for one product or surface. Teams working on several products keep them separated by using a project per product.

Actors

An actor is an AI assistant you create. It is defined by its instructions, its knowledge, and its capabilities:

• Instructions describe the actor's role and how it should behave.
• Knowledge is the reference material the actor can draw on, such as facts about your product or test credentials.
• Capabilities decide what the actor can do, for example browsing the web or reading email.

Under the hood, an actor drives a real browser and works through your product the way a person would, observing the page, taking an action, observing the result, and repeating until the task is done.

Checklists

A checklist is an automated verification workflow. You describe the steps you want checked in plain language, and an actor runs through them in a hosted browser. Each step is reported as passed or failed, with a screenshot and a short explanation.

Checklists are the fastest way to catch regressions in flows that are tedious to test by hand, such as signup, checkout, or onboarding.

Interviews

An interview is a structured conversation Stunt Double runs with your users. You define sections and questions, the platform conducts the interview, and it produces a report summarising what it learned. Interviews are useful for user research and feedback gathering at a scale that manual interviews cannot reach.

Workflows

A workflow chains steps into a single repeatable automation. Where a checklist verifies a flow, a workflow is meant to get something done end to end, and can combine several actions into one run.

Insights

As actors work, Stunt Double surfaces insights: notable observations drawn from a run, such as a confusing step, a broken link, or an unexpected result. Insights turn raw session recordings into something you can act on without watching every replay.

Recordings and replays

Everything an actor does is recorded. For any run you can replay the session, step through each action, and see the screenshot captured at that moment. This is what makes a failed checklist or a surprising interview answer easy to diagnose.

Next steps

Ready to automate this? See the API overview and the MCP server.