3 AGENTS · 3 WORKTREES · 1 MERGE
Controlled parallel development
Several AI coding agents work at once. You approve the final merge.
Five minutes to start
Install the CLI, sign in, connect one folder. Remaining steps stay visible on screen, and progress survives closing the tab.
Parallel work you can still see
Each ticket carries its agent and moves through TODO → IN_PROGRESS → REVIEW → DONE. Several agents run at once, and one screen still tells you where everything is.
Eight at once, on different models
Run Claude Code, Codex, and Antigravity in one workspace. Which agent is doing what, and how much is left, shows up as numbers.
They never step on each other's files
Every agent runs in an isolated git worktree. Parallel work never overwrites another agent's changes, and a conflict shows up as a conflict instead of being quietly flattened. Nothing merges until a human approves it.
feature/audit-panel
ahead 3 · behind 0 · no conflicts
fix/board-query
ahead 7 · behind 4 · 2 conflicts
archive/repo-connect
ahead 0 · behind 12 · merged
Read the code an agent wrote, immediately
Review the diff opened from the task branch without leaving the workspace. You never approve something you have not looked at.
A team leaves a record
Who opened what, and when a role changed, lands in the audit log. Member workload and worktree groups sit on the same screen.
What happened stays available
Finished work accumulates as records with its evidence. PRs, tickets, and diffs stay linked, so you can retrace why later.
How the orchestrator splits the work
Give it one goal and it decomposes the work into tickets and routes them to agents. The recording below is that decomposition and dispatch, captured from a real workspace.
Which way do you start
If you build alone, start with the Founders beta. If you are evaluating a team or company-wide rollout, a conversation is faster.