1.2 billion commits a week, still stuck in review: the four stages of agentic engineering
GitHub took more than 1.2 billion commits in a single week this October — 400% more than a year earlier, and more than the whole of 2025 combined. Now ask yourself: do you feel 400% more productive than last year? Writing software was never only "writing". Once AI accelerates the implementation step, the pressure moves somewhere else. A Faros AI study found that when a team goes from low to high…
In October, GitHub processed more than 1.2 billion commits, a 400% increase compared to the previous year and surpassing the entirety of 2025. The pressure of accelerated implementation due to AI adoption shifts the focus elsewhere. A Faros AI study revealed that as teams transition from low to high AI adoption, pull requests spend 441% longer in review.
GitHub has broken down a developer's work into four stages: Plan, Build, Verify, and Observe. Each stage examines the capabilities of agents and the necessary human oversight.
In the Plan stage, humans and agents must gather scattered context, assess trade-offs, and create an implementation plan. Utilizing tools like Copilot in Slack or Teams, the team can effectively collaborate, check assumptions live, and delegate tasks to Copilot for easier work.
During the Build stage, agents maintain progress on tasks even when the human is occupied. Session continuity allows continuing a Copilot session on a different device, and cloud sandbox execution ensures tasks run independently. Copilot keeps project knowledge and executes code in isolated environments without affecting other tasks or the user's machine.
In the Verify stage, humans must allocate their attention wisely to review changes. Stack smaller pull requests, employ code quality checks, and use AI scanning to identify security vulnerabilities or rule violations. Agentic Autofix and Agent Merge can also handle routine fixes and merge once verification passes.
Lastly, in the Observe stage, humans assess the deployed change and monitor for any issues or feedback. AI Scan and AI secret detection can proactively identify vulnerabilities and secrets, while Agent Merge can handle failed checks and merge conflicts.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.