🔍 Read the full analysis: AI Agents Start 4 In 10 Pull Requests On GitHub: Reviewers Adapt 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 2026 report says more than 40% of pull requests on its platform in September began with a detectable coding agent, up from 17% in March. The company also introduced tools for AI-assisted review and model routing, as human developers remain responsible for deciding what ships.
AI coding agents started more than 40% of GitHub pull requests in September, compared with 17% in March, according to the platform’s Octoverse 2026 report. GitHub said that amounted to 110 million agent-started pull requests in one month. The report also described tools for reviewing code produced with agent assistance.
GitHub defines an agent-started pull request as one whose first draft came from a coding agent the company can detect. It says the figure is a floor because its telemetry may miss some agent use. The report includes public and private repositories unless otherwise specified, but GitHub did not provide a public-versus-private breakdown for this measure. A pull request is a proposed code change that is reviewed before it is incorporated into a project.
The report says 90% of agent pull requests were merged within two weeks in September, up from 83% in February. By comparison, the merge rate for pull requests without a detected agent remained about 86%. These figures show that agent-started work is being merged at a high rate; they do not establish that the changes were correct, safe or free of later rework.
GitHub’s announcements at its October 28 Universe conference focus in part on review. The company introduced Copilot code review Max, which uses reviewers from different model providers and combines their feedback, and Copilot Fusion, a system that routes tasks to models. GitHub also said Copilot review approvals can count toward required approvals. Its estimated price per review is $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 Work Becomes the Constraint
The reported growth indicates that software teams may need to devote more time to inspecting code changes. If agents produce first drafts faster than people can review them, review capacity may limit how quickly changes are evaluated and merged. GitHub’s launch of review products and model-routing tools is part of its response to this shift.
The volume also has implications for budgets and oversight. Organizations may need to plan for review costs alongside compute and engineering time, and to decide which kinds of AI-generated feedback can count toward formal approval. Review totals and merge rates alone cannot show whether agent use improves software quality. Teams would need to track measures such as rework, reverted changes and incidents to judge those outcomes.
For individual developers, the report describes work that includes specifying tasks and reviewing generated code. The use of agents also raises questions about entry-level work. Small, introductory coding tasks have often provided opportunities for less-experienced developers to learn, and some of these tasks may be suitable for agents. The material does not quantify effects on hiring or training, so the longer-term impact remains uncertain.
Popular right now
Samsung 85-Inch QN70H Neo QLED 4K Smart TV (2026)
Ninja Flip Toaster Oven and Air Fryer, 8-in-1, Stainless Steel, SP151
Ninja CREAMi Scoop & Swirl 13-in-1 Ice Cream and Soft Serve Maker, NC701
As an affiliate, we earn on qualifying purchases.
How GitHub Defines Agent Work
The figures come from GitHub’s annual Octoverse report, which covers October 1, 2025, through September 30, 2026. GitHub’s wording matters: its core measure is a pull request that “started” with an agent’s first draft, not necessarily one written entirely by an agent. A person may revise the draft substantially and the pull request still counts under that definition.
The report also uses related labels such as “agent-generated,” “agent-authored” and “agent-opened” in chart captions and industry discussion. Those terms are not as precise as the report’s stated definition. GitHub recalculated 2024 and 2025 figures using updated definitions and a revised window, so older Octoverse totals are not directly comparable with this year’s series.
GitHub reported broader growth across its platform, including 1.3 billion merged pull requests during the 12 months to September, compared with 402 million in 2024 and 538 million in 2025. It also said 55% of monthly active Copilot users used agent features, up from 35% a year earlier. Those figures provide scale for the shift, but do not by themselves isolate how much platform-wide growth was caused by agents.
“A pull request whose first draft came from a coding agent that we can detect.”
— GitHub, in its Octoverse 2026 report
Quality and Coverage Still Unknown
GitHub’s 40%-plus share is not a complete count: the company says its detection misses some agent use, but has not quantified how much. It also did not publish a split between public and private repositories for this metric, making it difficult to tell how the adoption rate varies between open-source projects and company code.
The report’s merge figures do not establish code quality. GitHub said agent pull requests were more likely to be merged within two weeks than other pull requests, but merging is not proof of correctness. The supplied report details do not provide comparative rates for defects, security problems, reverts or follow-up maintenance. Nor do they show whether human review time has risen or how teams are changing their review practices.
It is also not clear how widely teams will rely on Copilot approvals when those approvals can count toward required reviews, or how the new tools’ estimated per-review charges will affect actual spending. The definition of “started” does not distinguish a lightly edited agent draft from one substantially rewritten by a human.
Teams Will Test Review Rules
Organizations using coding agents will need to set review rules and monitor results. That includes deciding when AI review may count toward approval, who must sign off, and how to measure whether faster drafting leads to more rework or incidents. GitHub’s announced per-review estimates give teams a starting point for budgeting, but actual costs will also depend on usage and associated Actions minutes.
GitHub’s report offers an initial platform-wide snapshot, not a settled account of the effect on software quality or developer jobs. Further evidence will be needed to show how adoption changes across repositories and whether review practices keep pace. Quality, accountability and reviewer workload are among the issues to monitor as agent-started pull requests become a larger share of code proposals.
Key Questions
What does it mean that an AI agent started a pull request?
GitHub counts a pull request when its first draft came from a coding agent the company can detect. A human can revise that draft and it still counts under the report’s definition.
Is the 40% figure a complete count of agent-written pull requests?
No. GitHub says the figure is a floor because some agent activity may not be detected. It has not provided an estimate of how much use is missing.
Does the high merge rate mean agent code is better?
No. GitHub reported that 90% of detected agent pull requests were merged within two weeks in September, but merge rates do not measure correctness or long-term quality. The report details provided do not compare defects, reverts or incidents.
How much does GitHub estimate AI code review costs?
GitHub estimates an average review at $0.25 for Lite, $1.50 for Balanced and $10 for Max. GitHub Actions minutes are additional, and the estimates do not establish what every team will spend.
Who is responsible for code an agent produces?
GitHub says developers remain accountable for what ships. Its stated controls also prevent its cloud agent from approving or merging its own pull request, and prevent the person who requested a change from approving it.
Source: ThorstenMeyerAI.com
