GitHub outlines agentic engineering workflow as review times stretch with AI adoption
GitHub argues that faster AI-assisted coding shifts developer effort toward planning, review and operations, and describes a four-stage workflow using Copilot across Plan, Build, Verify and Observe.
In plain words
GitHub, the code-hosting company, says AI agents can now help with much more than writing code. It describes a workflow where agents assist in planning work, running tasks, reviewing changes and investigating problems after release. The company also cites a study finding that teams with heavy AI use saw pull requests waiting far longer for review.
What was announced
The guide, published on the GitHub Blog, is anchored by a striking figure: GitHub recorded more than 1.2 billion commits in a single week this October, over 400% more than a year earlier and more than in all of 2025. The company asks whether developers feel correspondingly more productive, and cites a Faros AI study finding that pull requests spent 441% longer in review when teams moved from low to high AI adoption. GitHub's answer is a workflow spanning four stages—Plan, Build, Verify and Observe—where agents help gather scattered requirements, run tasks in cloud sandboxes, and keep work moving across devices.
On the review bottleneck, GitHub recommends smaller, stacked pull requests, automated code review with adjustable depth, and tools such as GitHub Code Quality, AI Scan and Agentic Autofix to clear routine checks before a human looks at the diff. For post-release work, it describes Copilot drawing on observability data through connectors to investigate incidents and draft follow-up issues. The piece closes with advice to start with one frustrating task rather than overhauling a whole process, and notes the topic was presented at GitHub Universe 2026.
What it does not show
- The 441% review-time figure comes from a Faros AI study cited by GitHub; the underlying study is not included in the source, so its methodology and scope cannot be verified.
- The commit statistics and workflow descriptions are GitHub's own claims and product descriptions, not independently confirmed.
- Several features, such as package firewall in public preview, may have limited availability; the source does not specify general availability for all tools mentioned.
For security and IT teams
- Scope
- The source is a GitHub Blog guide describing an agentic engineering workflow across Plan, Build, Verify and Observe stages, presented at GitHub Universe 2026.
- Access
- GitHub package firewall is described as currently in public preview for npm, and enterprise managed settings carry policies across Copilot surfaces.
Sources
Source-support review completed 13 Nov 2026. This story was drafted with AI assistance and checked against retained source passages. Automated checking is not independent verification.
Updates and corrections
No corrections so far.
First published 13 Nov 2026
New to the subject? AI from the Ground Up starts from the beginning.