🔍 Read the full analysis: The Rise Of AI-Started Pull Requests Is Shifting Work To GitHub Reviewers on ThorstenMeyerAI.com
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
TL;DR
GitHub’s Octoverse report says more than 40% of pull requests on its platform in September 2026 began with a detectable coding agent, up from 17% in March. The report also says 90% of those agent-started requests were merged within two weeks, but merging does not establish that code was correct. GitHub’s new review tools and pricing reflect a shift toward checking, routing and approving AI-generated work.
More than 40% of pull requests on GitHub in September began with a first draft from a detectable coding agent, the company said in its Octoverse 2026 report, up from 17% in March. The increase is shifting work toward the people and tools that inspect proposed changes: GitHub introduced new AI review features this week, while saying developers remain accountable for code that ships.
GitHub counted 110 million agent-started pull requests in September. It defines one as a pull request whose first draft came from a coding agent the company can detect. GitHub says the number is a floor because some agent use may not be detected. The measure includes public and private repositories, but the report does not provide a separate breakdown for them.
The company reported that 90% of detected agent-started pull requests were merged within two weeks in September, compared with 83% in February. Pull requests without a detected agent had a merge rate of about 86%. Those figures describe whether changes were merged, not whether the code was correct, secure or free of defects.
GitHub announced Copilot code review Max, which uses several AI reviewers from different model providers and combines their feedback, and Copilot Fusion, a system that routes work to models for different steps. It also said Copilot approvals can count toward required approvals. GitHub estimates reviews cost about $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 Workload
The figures point to a change in where software-development time is spent. When an agent produces a proposed change quickly, reviewers still need to understand its purpose, inspect the diff, check interactions across files and decide whether it is safe to merge. GitHub’s report says 87% of agent-started pull requests changed more than one file, compared with 79% of requests without a detected agent. That can increase the amount of context a reviewer must evaluate, though the statistic alone does not establish that each change is harder to review.
For engineering teams, the practical question is not only how many pull requests agents create, but whether review capacity and quality controls can keep pace. Organizations may need to budget for reviewer time and paid review usage, set rules for when AI-generated approvals count, and track outcomes such as rework, reverts and incidents. Pull-request volume or merge speed alone cannot show whether the code improved productivity.
The shift also affects developers’ work and training. Drafting routine changes may become less central, while writing clear tasks and critically checking generated code become more important. If agents take on small starter tasks, managers may need other ways to give junior developers practice in understanding codebases and making changes under review.
Popular right now
Breville BES870XL Barista Express Espresso Machine, Brushed Stainless Steel
De'Longhi Rivelia Fully Automatic Espresso Machine with Milk Frother, White
Roborock Qrevo CurvX Robot Vacuum and Mop
As an affiliate, we earn on qualifying purchases.
How GitHub Defines Agent Work
The report covers the year from October 1, 2025, to September 30, 2026. GitHub says agent-started requests are identified through signals it can detect; it does not claim to capture every instance of AI assistance. A human may also revise an agent’s initial draft and the request will still count as agent-started. The company’s wording varies across the report, but “started with an agent’s first draft” is the most precise description of the metric.
GitHub says it recalculated 2024 and 2025 figures using the revised definitions and reporting window, so comparisons with older Octoverse reports are not direct. The broader platform also saw substantial activity: GitHub reported 1.3 billion merged pull requests during the year, up from 538 million in 2025. Its report describes agent use as growing alongside that platform-wide increase, rather than isolating how much of the overall change agents caused.
GitHub also said that 55% of Copilot monthly active users used agent features, compared with 35% a year earlier, and that use of its cloud agent grew 230% year over year. These are company-reported usage measures. They do not on their own show the quality of the resulting code or the effect on development costs.
“An agent-started pull request is “a pull request whose first draft came from a coding agent that we can detect.””
— GitHub, in its Octoverse report
What the Merge Numbers Cannot Show
The report does not show how many agent-started pull requests led to defects, security problems, reverts or extra review work. A merge is not a quality measure, and GitHub’s reported figures do not establish that agents caused the rise in total merged requests. The report also gives no public-versus-private split for agent-started pull requests, leaving the distribution across open-source and company repositories unclear.
Because GitHub counts detectable signals, the actual share of requests that began with agent work may differ from the published figure. The company characterizes its measure as a floor, but does not quantify how much activity is missed. It is also not clear from the source material how often organizations will allow Copilot approvals to satisfy their own required-review policies, or how the new tools will affect review accuracy and total costs in practice.
Teams Must Test Review Policies
GitHub’s new review products put pricing, model routing and approval settings into the workflow as agent use grows. Teams will need to decide which changes qualify for automated review, when a human must approve, and how to monitor the cost of reviews alongside GitHub Actions usage. The source material does not give a timeline for how widely the new features will be adopted.
The next useful evidence will be operational: whether AI review reduces reviewer workload without increasing missed defects, and whether agent-generated changes require more rework after merging. For now, GitHub’s report documents a fast increase in detected agent-started pull requests, while the effects on code quality and engineering productivity remain unsettled.
Key Questions
What does GitHub mean by an agent-started pull request?
GitHub defines it as a pull request whose first draft came from a coding agent the company can detect. A human can revise the draft and it still counts under that definition.
How many pull requests did agents start?
GitHub reported 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 be missed.
Does a high merge rate mean agent code is reliable?
No. GitHub said 90% of detected agent-started requests were merged within two weeks, but merging does not establish correctness or code quality. The report does not provide defect or rework rates for those changes.
How much does Copilot code review cost?
GitHub estimates an average review at $0.25 for Lite, $1.50 for Balanced and $10 for Max. GitHub Actions minutes are an additional cost.
Who is responsible for code that an agent helps produce?
GitHub says developers remain accountable for what ships. It also says its cloud agent cannot approve or merge its own pull requests, and the person who requested a change cannot approve it.
Source: ThorstenMeyerAI.com
