GitHub adds AI approval controls and speeds up code review process

GitHub announced new AI‑agent features for GitHub Issues, introducing three controls – Approvals, Confidence, and Rationale – that let AI propose label changes, assignee updates and issue closures while displaying a high/medium/low confidence rating and the reasoning behind each suggestion. The update, currently in public preview, requires a human to review and approve lower‑confidence actions, enabling a "human‑in‑the‑loop" workflow for safer AI deployment.

At the same time, industry focus on accelerating software delivery is highlighted by Autodesk lead software engineer Gleb Sinani, who reduced his team's code‑review cycle from three days to under one business day. Sinani achieved this by redefining pull‑request flow rules, clarifying ownership, improving notification timing, and restructuring pipelines into modular CI/CD paths, thereby cutting infrastructure complexity and speeding time‑to‑market.

These initiatives reflect a broader move among technology firms to combine process redesign with AI assistance to shorten development cycles and increase release reliability.