Close the chat.
Keep work moving.
Your planning session can end while approved work keeps running. Come back to the results, or a blocker that needs your decision.
BacklogGo is an out-of-the-box, self-hosted software factory. Shape the plan with your coding agent, then keep approved work moving after you close the chat.
Two concurrent tasks free. No website account required. AI usage and infrastructure are separate.
Plan in your repo with your own coding agent. Review the scope and assumptions, then approve the plan and let your agent create the board tasks.
The worker implements ready tasks, runs relevant checks and saves the results for review.
A separate reviewer assesses the change and verification results. Work can return for correction.
Final quality approval gates delivery. Choose a human-reviewed route when you want the final merge in your hands.
When work hits a blocker, its explanation and evidence stay with it, ready for your decision.
Talk to customers. Shape the next feature. Give approved work a place to run while you handle the rest of building a product.
Your planning session can end while approved work keeps running. Come back to the results, or a blocker that needs your decision.
Run independent tasks concurrently, in one repository or across projects. Add capacity when your workload and infrastructure justify it.
Saved progress helps work resume after a restart. If it cannot continue safely, the task tells you what needs attention.
Try parallel work on a real project. Add capacity when you need more independent tasks moving at once.
BacklogGo runs on your own infrastructure. The website account is for purchasing and managing a license, rather than hosting your projects.
Use Codex or compatible models through the Pi agent harness. The current Pi integration connects to OpenAI-compatible model endpoints. Configure your model access during setup.
No. Your license is a signed file stored with your installation. BacklogGo verifies it locally, fully offline. There are no cloud license checks or periodic online check-ins.
No. Bring your own model access through Codex or Pi. Model usage and infrastructure costs are separate from your BacklogGo license.
No. Your configured GitHub or GitLab board remains the source of work and status.
Yes. Delivery follows your configured policy. A human-reviewed route can leave the request for you to review and merge.
Two free concurrent tasks. No website account required.