Parallax
Local-first AI orchestrator for software development tasks.
Start new thread
Maxi Gimenez

Parallax - Local-first AI orchestrator for software development tasks.

by
Parallax
•
Parallax is a local-first AI orchestrator for software development tasks. It pulls work from Linear or GitHub Issues, creates isolated worktrees, generates a plan first, waits for approval, then executes changes and opens or updates the related branch and PR while keeping control on your machine.

Add a comment

Replies

Best
Maxi Gimenez
Maker
📌
Hey Product Hunt! I’m excited to share Parallax with you. I built it because I wanted an AI coding agent that fits into a real dev workflow, not one that disappears into a black box. Parallax runs locally, turns Linear or GitHub issues into tasks, creates a plan first, waits for approval, then works in isolated branches and opens a reviewable PR. The big idea is simple: let AI help with real software work while you stay in control of planning, execution, and review. If that sounds useful, I’d love for you to check it out and tell me what feels promising, confusing, or missing. I’ll be around all day and would really appreciate the feedback.
Sviatoslav Dvoretskii

The plan-first-then-execute workflow with explicit approval gates is exactly the right trust model for AI coding agents — pulling issues directly from Linear or GitHub, creating isolated worktrees, and opening reviewable PRs means Parallax slots into an existing dev process rather than requiring teams to adopt a new one. Running everything locally with full control on your machine is a strong differentiator from cloud-based coding agents; how does Parallax handle issues that require changes across multiple repos or services — can it coordinate work across different worktrees, or is it scoped to single-repo tasks?

Maxi Gimenez
Maker

@svyat_dvoretski Great question, and thanks for calling out the workflow design.

Right now Parallax is intentionally scoped to one repository per task (a 1:1 relationship between a task and a repo). That constraint is mostly about keeping execution predictable and reviewable, the agent plans, creates an isolated worktree, and opens a PR in the same repo the task came from.

Maxi Gimenez
Maker

🚀 Big News: Parallax is Now Open Source!

I'm excited to share that Parallax, it’s officially open source.

Parallax started as an idea to make working AI workflows simpler and more transparent.

By making Parallax open source, anyone can:
✨ Explore how it works
🔧 Contribute improvements
🚀 Build new ideas on top of it


I’m really excited to see what the community builds, and I’d love your feedback, contributions, and ideas.

If you find Parallax interesting, I’d really appreciate your support on Product Hunt today 🙌

Abhinav Ramesh

Local-first AI orchestration is an underrated design choice — no latency spikes, no data leaving the machine. How does it handle tasks that need external API calls while staying local?

Maxi Gimenez
Maker

@abhinavramesh It's handled as a normal HTTP call. If it fails for any reason, the response is processed, and the task is canceled. In this case, the error is logged and displayed on the UI dashboard or when running "parallax status." The user can retry the task at any time.