🔍 Read the full analysis: More AI-Started Pull Requests On GitHub Mean A Different Review Workflow on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
GitHub’s Octoverse report says more than 40% of pull requests on the platform in September 2026 began with a coding agent’s first draft, up from 17% in March. The increase is changing the review workflow: GitHub introduced new AI review and model-routing tools, while saying human developers remain accountable for code that ships.
More than 40% of pull requests on GitHub in September 2026 began with a coding agent’s first draft, up from 17% in March, according to the platform’s 2026 Octoverse report. The shift means teams are handling far more agent-started code and places greater weight on review, approval and accountability—not just how quickly code can be written.
GitHub says the September share represents 110 million agent-started pull requests in one month, roughly six times the volume in March. The company defines an agent-started pull request as one whose first draft came from a coding agent GitHub can detect. It says the count is a floor, because some agent use may not be detected. The report’s figures include public and private repositories unless otherwise specified; GitHub does not provide a public-versus-private breakdown for this measure.
GitHub also reports that 90% of detected agent pull requests in September were merged within two weeks, compared with 83% in February. Pull requests without a detected agent remained at about 86%. These are merge rates, not measures of whether changes were correct, safe or free of later rework. Agent-started changes also touched more files: 87% affected multiple files, compared with 79% of pull requests without a detected agent.
The company’s product announcements this week center on review. Copilot code review Max uses several AI reviewers from different model providers and combines their findings. GitHub also made Copilot approvals eligible to count toward required approvals and released Copilot Fusion, which routes work to models for different steps. GitHub estimates an average review costs $0.25 in Lite mode, $1.50 in Balanced and $10 in Max, with GitHub Actions minutes charged separately.
Infographic 01 · The agent share
Four in ten pull requests now start with an agent.
GitHub’s own telemetry, March to September 2026. The first draft comes from an agent; review and merge stay with people.
Merged within two weeks
Share of pull requests merged within 14 days
Merged fast is not the same as correct. GitHub publishes no revert or defect rate.
Bigger changes to review
Share of pull requests that change more than one file
More files per change means more for a reviewer to check.
What GitHub counts: an “agent-started pull request” is one “whose first draft came from a coding agent that we can detect.” Undetected agents are missing, so GitHub calls the figure a floor. Metrics cover public and private repositories unless noted.
Source: GitHub Octoverse 2026, 28 Oct 2026 (text, chart captions, glossary and methodology). All figures are GitHub’s own.
Powered by Thorsten Meyer AI
Review Becomes the New Constraint
When agents can prepare code drafts quickly, the scarce resource shifts toward people who can read, test and approve changes. A higher volume of pull requests can put pressure on reviewer capacity, particularly when changes touch several files and require a developer to understand how the pieces fit together. The reported rise in merges shows that agent-started work is entering repositories; it does not establish that the work is correct or that teams are saving time overall.
The cost of review also becomes a planning decision. GitHub’s estimated per-review prices vary widely by mode, and the company says Actions minutes add to the bill. Organizations may need to decide which changes warrant a more involved AI review, when a Copilot approval can satisfy a repository rule, and where a human review remains necessary. Teams should track rework, reversions and incidents alongside output volume and review spend to judge the results.
The change may also affect how developers learn. Routine, smaller tasks have often given less experienced engineers practice in understanding a codebase and proposing changes. If agents take on more of that drafting, teams will need to make sure junior developers still gain experience writing, reviewing and debugging code. That is a possible workforce implication, not an outcome established by GitHub’s platform figures.
Products Worth Considering
Bose QuietComfort Wireless Noise-Cancelling Over-Ear Headphones, Black
Bose QuietComfort Wireless Bluetooth Headphones with Active Noise Cancellation, Mic, USB‑C Charging, Deep Bass, 24‑Hour Battery – White Smok
Logitech MX Mechanical Wireless Illuminated Keyboard Tactile (Graphite)
As an affiliate, we earn on qualifying purchases.
How GitHub Defines Agent-Started
The Octoverse report is GitHub’s annual review of activity on its platform and covers October 1, 2025, through September 30, 2026. The company recalculated its 2024 and 2025 figures using revised definitions and a new measurement window, so earlier reports are not directly comparable. The report’s exact phrase for the headline measure is that a pull request “started with an agent’s first draft”; other passages use terms such as “agent-generated” or “agent-authored.” The definition, rather than those looser labels, is the clearest guide to what is counted.
The growth comes amid a broader increase in GitHub activity. The report says developers made 17.9 billion code pushes during the year and that merged pull requests reached 1.3 billion, up from 538 million in 2025. GitHub also says 55% of monthly active Copilot users now use agent features, compared with 35% a year earlier, while use of its cloud agent grew 230% year over year. These platform-wide totals provide scale but do not, on their own, show whether agents caused the overall growth.
GitHub says its cloud agent cannot approve or merge its own pull requests, and the person who requested a change cannot approve it either. The company’s stated policy keeps responsibility for shipped code with developers, even as AI tools take on more drafting and review work.
“A pull request whose first draft came from a coding agent that we can detect.”
— GitHub, in the Octoverse report’s definition of an agent-started pull request
What the Platform Data Cannot Show
The figures do not reveal how many changes were correct, how much human editing each draft received, or whether merged agent-started pull requests later required rework, were reverted or caused incidents. A merge is an outcome in GitHub’s workflow data, not an independent quality assessment. The report also offers no split between public and private repositories for the agent share, and its detection limits mean the true proportion is unknown.
It is also unclear how much review time the new tools save in practice, how often teams will use each paid review mode, or what the added review costs will be across organizations. GitHub’s figures describe activity on its platform; they do not establish that every company will see the same adoption rate or productivity effects.
Teams Set Their Review Rules
The next step for organizations is deciding how agent-started changes fit existing repository policies: which AI findings trigger human review, whether Copilot approvals count for particular projects, and how review costs are budgeted. Teams will need to compare the tools’ costs with measured outcomes such as review time, rework and production issues, rather than relying on pull request or merge totals alone.
GitHub’s report covers activity through September 30, 2026, while its review and routing products have just been announced or released. The available material does not give adoption figures for those new features or a timeline for when their effects may be assessed. Further evidence from teams using the tools will be needed to show whether AI review keeps pace with the rising volume of agent-started code.
Key Questions
What does GitHub mean by an agent-started pull request?
It is a pull request whose first draft came from a coding agent that GitHub can detect. The count does not require the agent to have written every line that appears in the final request.
How many pull requests did agents start in September?
GitHub reports 110 million in September 2026, representing more than 40% of pull requests on the platform that month. The company says its detection-based count is a floor because some agent use may go unrecorded.
Does the 90% merge rate show that agent-written code is reliable?
No. GitHub says 90% of detected agent pull requests were merged within two weeks in September. Merge rate is not a code-quality measure, and the report does not establish how many changes later needed rework or caused problems.
How much does GitHub’s AI code review cost?
GitHub estimates an average review at $0.25 in Lite mode, $1.50 in Balanced mode and $10 in Max mode. GitHub Actions minutes are additional, and actual spending will depend on usage.
Who is responsible for agent-written code that ships?
GitHub says developers remain accountable. Its cloud agent cannot approve or merge its own pull requests, and the person who requested a change cannot approve that request either.
Source: ThorstenMeyerAI.com
