Product overview

One system for the request, the run, and the decision.

M1 Project is a proposed shared workspace for product and engineering teams. People shape and own the work; configured agents execute bounded tasks on connected runtimes; evidence returns to the same record for review.

01 · Capture

Start with a task, not a disposable prompt

A task holds the outcome, constraints, priority, type, status, project, sprint, discussion, and acceptance criteria. That shared record gives people and future agent runs the same starting point.

  • ↗Separate human ownership from agent execution
  • ↗Keep changes, comments, and run evidence together
  • ↗Move work through typed, team-defined workflows

02 · Execute

Route bounded work to a reusable agent

Configure a role once with instructions, skills, repository scope, runtime, environment, concurrency, timeout, and directory mode. Teammates can then assign that operating pattern without rebuilding a prompt.

  • ↗Choose a connected computer and available runtime
  • ↗Use isolated worktrees for parallel repository tasks
  • ↗Surface waiting, running, blocked, failed, and completed states

03 · Decide

Make human review a visible stage

An agent result is a candidate, not an automatic release. Reviewers inspect the summary, changed files, checks, open questions, and trace before accepting, asking for changes, merging, or releasing.

  • ↗No implied automatic merge or deployment
  • ↗Acceptance stays attributable to a person
  • ↗Follow-up comments continue the same task context

Pre-release · No instant signup or payment

See whether M1 Project fits your team’s operating model.

Request a walkthrough