🔍 Read the full analysis: The GitHub Pull Request Shift: AI Starts 4 In 10, Review Picks Up on ThorstenMeyerAI.com
Get smart everyday buys 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 in September 2026 began with a detectable coding agent, up from 17% in March. The company also introduced AI review and model-routing tools as it turns attention to evaluating agent-written changes; its figures do not show whether merged code was correct.
More than 40% of pull requests on GitHub in September 2026 began with a detectable coding agent, up from 17% in March, according to the platform’s Octoverse report released October 28. GitHub says that represents 110 million agent-started pull requests in one month, and its accompanying product launches focus on reviewing and approving code as agents take on more of the initial drafting.
GitHub defines an agent-started pull request as one whose first draft came from a coding agent the company can detect. It describes the count as a floor because some agent use may not be detected. A pull request is a proposed code change that is reviewed before it is incorporated into a project. The figure includes public and private repositories under the report’s stated methodology, but GitHub did not provide a split between them.
The report says 90% of agent-started pull requests were merged within two weeks in September, compared with 83% in February. Pull requests without a detected agent remained at about 86% over the period cited. Those figures measure whether changes were merged, not whether they were correct, safe or beneficial after release.
GitHub announced Copilot code review Max, which draws on several AI reviewers from different model providers and combines their reviews. It also said Copilot review approvals can count toward required approvals and introduced Copilot Fusion, a router that selects models for different steps. GitHub estimates an average review costs $0.25 for Lite, $1.50 for Balanced and $10 for 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
The figures suggest that code drafting is no longer the only, or necessarily the slowest, part of a development workflow. When agents generate more proposed changes, teams still need people or tools to inspect them, decide whether they meet requirements and take responsibility for accepting them. Review capacity and reviewer attention can become constraints even when producing a first draft takes less time.
The reported merge rate may look like evidence that agent-written changes are being accepted quickly, but it is not a quality measure. A merged pull request can still require substantial rework, cause a later failure or be reverted. Companies will need to compare adoption with outcomes such as rework, reversions and production incidents, rather than relying on pull-request counts alone.
Pricing also turns review into a budget decision. The listed costs differ substantially by review tier, and the source says they sit alongside charges for Actions minutes. Teams adopting the tools will need to decide where a lower-cost review is sufficient, when a more involved review is warranted, and whether an AI-generated approval should count toward a project’s formal approval requirements.
Products Worth Considering
Logitech MX Mechanical Wireless Illuminated Keyboard Tactile (Graphite)
Dell UltraSharp U2725QE 27-Inch 4K UHD IPS Black Monitor
Dell 24 Plus Monitor S2425HSM – 23.8 FHD (1920×1080) 144Hz 1ms, 2 x 3W Speakers, HDMI, Height/Tilt/Pivot/Swivel Adjustability, AMD FreeSync
As an affiliate, we earn on qualifying purchases.
How GitHub Defines Agent Work
The Octoverse 2026 report covers the year from October 1, 2025, through September 30, 2026. GitHub recalculated its 2024 and 2025 figures using updated definitions and a different reporting window, so the company cautions against direct comparisons with older Octoverse reports. For this year’s stated comparison, the agent-started share rose from 17% in March to more than 40% in September; GitHub also says September’s volume was about six times March’s.
GitHub’s terminology is not fully consistent across the report: it uses “agent-started,” “agent-generated,” “agent-authored” and “agent-opened” in different places. Its definition centers on the first draft, not on how much a person later changes. That distinction matters: a pull request can count as agent-started even if a developer substantially rewrites the proposal before submitting it.
The report describes broader growth in activity on the platform, including 1.3 billion merged pull requests during the year, up from 538 million in 2025. GitHub also says 55% of Copilot monthly active users now use agent features, compared with 35% a year earlier. These platform-wide figures provide context, but they do not establish that agents alone caused the increase in merged changes.
“A pull request whose first draft came from a coding agent that we can detect.”
— GitHub, in the Octoverse 2026 report
What the Adoption Figures Cannot Show
GitHub has not published a public-versus-private repository breakdown for the agent-started share, nor a detailed account of which agents its detection covers. Because undetected use is excluded, the reported 40%-plus figure is not a complete census of AI-assisted work. The first-draft definition also does not measure how much human editing followed.
The report’s merge statistics do not establish code quality, security or downstream performance. It is unclear from the supplied figures how often merged agent-started changes later required substantial repair, were reverted or contributed to incidents. GitHub’s product announcement likewise does not provide comparative evidence that its review tools reduce those outcomes.
Some labels in the report vary, and the revised historical calculations limit comparisons with previous Octoverse editions. The data therefore support a conclusion about rapid growth in detected agent-started work, but not a precise estimate of all AI coding activity or its net effect on software quality and productivity.
How Teams Will Test AI Review
GitHub has announced the new review and routing features, but the report does not provide a timeline for broader availability or evidence from controlled comparisons of their performance. Teams considering them will need to track review costs, time to merge and the quality of changes after release, while accounting for the separate cost of Actions minutes.
For organizations, the immediate decisions include who can approve agent-written changes, whether Copilot approvals count toward required checks and which changes need additional human scrutiny. GitHub says its cloud agent cannot approve or merge its own pull requests, and the person who requested the change cannot approve it either. How companies set rules beyond those stated product restrictions remains a matter for their own review policies.
For developers, the shift makes clear task descriptions and careful reading of diffs more consequential. The next useful evidence will be whether agent adoption leads to durable gains without increasing defects, rework or review burden—outcomes that the current adoption and merge figures do not resolve.
Key Questions
What does GitHub mean by an agent-started pull request?
GitHub counts a pull request when its first draft came from a coding agent the company can detect. Later human edits do not remove it from that category.
Does a high merge rate mean the agent-written code is good?
No. GitHub reports that 90% of agent-started pull requests were merged within two weeks in September, but merging does not prove correctness or quality. The figures do not report subsequent rework, reversions or incidents.
How much does Copilot code review cost?
GitHub estimates average costs of $0.25 for Lite, $1.50 for Balanced and $10 for Max. GitHub Actions minutes are additional, according to the source material.
Does GitHub’s percentage cover all AI-assisted pull requests?
No. The measure covers agents GitHub can detect. The company calls it a floor because some agent use may not be detected, and it has not published a repository-type breakdown for this figure.
Source: ThorstenMeyerAI.com
