Intent by Augment Code
A workspace for running multiple AI coding agents on the same codebase at once, using isolated git worktrees and a shared living spec to avoid conflicts.
🔗 Visit Intent by Augment Code
Description
Running one AI coding assistant is easy; running several at the same time on the same project usually ends in merge conflicts and confusion about who changed what. Intent is built around that specific problem — instead of one agent working alone in your editor, it lets a whole team of agents work in parallel on the same codebase without stepping on each other.
Intent, from Augment Code, is a macOS desktop app (public beta) that coordinates multiple AI agents through isolated git worktrees so each one edits its own copy of the repo, plus a shared "living spec" document that keeps humans and agents aligned on what's actually being built. Each task gets a coordinator agent that proposes a plan, implementer agents that execute the work in parallel worktrees, and a verifier that checks the result before it reaches human review. It supports bring-your-own-agent (BYOA) setups, includes a consolidated terminal and full git integration, and is currently free during the public beta on macOS, with Windows support on a waitlist and final pricing not yet announced.
💬 Our review
The short version: if you're already running multiple AI coding agents on a project and constantly untangling their conflicting changes by hand, Intent's worktree-based coordination model solves a real, specific pain point — but it's a beta product from a young space, so expect rough edges.
Against Claude Code and OpenCode, which run one agent at a time in your terminal, Intent's whole pitch is orchestration: a coordinator plans the work, several implementers execute it in isolated git worktrees so they can't clobber each other's changes, and a verifier checks the result before you ever see it. Google's Antigravity is the closest conceptual competitor, tackling the same multi-agent-on-one-repo problem, though Intent leans more on git-native isolation while Antigravity leans more on its own workspace abstraction. Being free during a macOS-only public beta makes it a low-risk trial for teams already juggling multiple agents, but the lack of Windows support and unannounced pricing mean it's not yet a tool to build a workflow around long-term — treat it as a promising bet on where AI-assisted development is heading rather than a finished product.
📊 Global score
🤖 AI-enriched data
Gratuit pendant la bêta publique macOS. Windows en liste d'attente. Modèle tarifaire final non annoncé.
Pros
Orchestration multi-agents sans conflits via worktrees git isolés
Spec vivante partagée qui garde humains et agents alignés
Coordonnateur qui propose un plan avant l'exécution parallèle, plus un vérificateur avant revue humaine
Support BYOA (apportez votre propre agent) et terminal/git intégrés
Cons
Bêta publique : pas de modèle tarifaire final annoncé
Windows pas encore disponible (liste d'attente)
Peu de documentation publique sur les limites du nombre d'agents supportés