🔍 Read the full analysis: Reviewing Code Is The Growing Challenge As AI Agents Start GitHub PRs on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
GitHub’s Octoverse report says more than 40% of pull requests on the platform in September 2026 started with a detectable coding agent, up from 17% in March. The company introduced AI review and model-routing tools as code review becomes a larger part of the development workload, while saying developers remain accountable for what ships.
More than 40% of GitHub pull requests in September 2026 began with a coding agent’s first draft, up from 17% in March, according to GitHub’s Octoverse report released October 28. The increase shifts pressure toward reviewing and approving code: GitHub introduced new AI review and model-routing tools this week, while maintaining that developers remain accountable for code that ships.
GitHub counted 110 million agent-started pull requests in September, about six times the March volume, according to the report. It defines an agent-started pull request as one whose first draft came from a coding agent the company can detect. GitHub describes the count as a floor because some agent use may not be detected. The measure includes public and private repositories, and the report does not provide a breakdown between them.
The report says 90% of detected agent pull requests were merged within two weeks in September, compared with 83% in February. Pull requests without a detected agent held at about 86%. A merge indicates that a proposed change was accepted into a codebase; it does not by itself show that the change was correct, safe or free of later rework.
GitHub’s new offerings focus on review and workflow. Copilot code review Max combines reviews from several AI models into one, while Copilot Fusion routes different tasks to models selected for each step. GitHub also made Copilot approvals eligible to count toward required approvals. The company estimates a review costs an average of $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 Capacity Becomes the Constraint
When agents produce first drafts faster, the workload does not disappear; it shifts toward checking changes, evaluating risks and deciding whether to merge. The growth in agent-started pull requests makes review capacity a practical concern for teams, particularly where reviewers must understand changes across multiple files or check how they interact with existing systems.
GitHub reports that 87% of agent-started pull requests changed more than one file, compared with 79% of pull requests without a detected agent. That difference does not establish that agent changes are harder to review or more error-prone, but it suggests some reviewers may need to assess a broader set of edits in a single proposal.
The cost estimates also make review a budgeting question. Organizations considering AI review will need to weigh the per-review price against reviewer time and the consequences of missed defects. Useful measures include rework, reversions and production incidents, not just how many pull requests agents open or how many are merged. Those outcomes are not established by the merge-rate figures in the report.
For developers, the change puts more emphasis on writing precise tasks and critically reading diffs. If routine starter tasks are increasingly assigned to agents, teams may also need to consider how junior developers gain experience with code changes and review. The report does not measure that effect, so its scale remains unclear.
Products Worth Considering

Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite

Logitech MX Mechanical Mini for Mac Wireless Keyboard, Space Gray

Dell UltraSharp U2424H 24-Inch FHD Monitor
As an affiliate, we earn on qualifying purchases.
How GitHub Counts Agent Work
The Octoverse report is GitHub’s annual review of activity on its platform. Its 2026 edition covers October 1, 2025, through September 30, 2026. GitHub’s measure is based on detectable signals and counts a pull request as agent-started when an agent supplied its first draft, even if a person later revised the work. The figure therefore does not mean every line in those pull requests was written by an agent.
GitHub uses several related descriptions in the report, including “agent-started,” “agent-generated” and “agent-authored.” Its definition centers on the first draft, making agent-started the most precise description of what the headline figure measures. GitHub also recalculated figures for 2024 and 2025 using updated definitions and the reporting window, so the report says older Octoverse numbers are not directly comparable.
The company reported wider growth in platform activity during the same period: 1.3 billion pull requests were merged, up from 538 million in 2025 and 402 million in 2024. Those totals cover GitHub activity overall and should not be read as agent-only figures. GitHub also said 55% of Copilot’s monthly active users now use agent features, compared with 35% a year earlier.
“A pull request whose first draft came from a coding agent that we can detect.”
— GitHub, in its Octoverse report
Limits of the Agent Metrics
The reported share is not a census of all AI-assisted coding: GitHub counts agents it can detect and says the result is a floor. The company has not published a public-versus-private split for the statistic, nor a measure of how much agent-written code people changed before submitting a pull request.
The rise in merges also does not establish code quality. The report’s 90% two-week merge rate describes acceptance, not defect rates, security outcomes or long-term maintenance costs. The source material does not provide a comparable measure of later rework, reversions or incidents for agent and non-agent pull requests.
It is also not clear how widely teams will use Copilot approvals to satisfy required review rules, or how organizations will set limits on automated approval. 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 report does not specify how every company’s own review policies will interact with those controls.
Teams Set Review Rules and Budgets
GitHub’s immediate next step is making its review and routing features available to users, with the stated per-review estimates providing one basis for teams to plan usage. Companies adopting them will need to decide which changes may receive AI review, when human approval is required and how costs are tracked. GitHub Actions minutes add to the listed review prices.
The next useful evidence will come from measurements beyond activity and merge rates: whether AI review catches defects, how often human reviewers revise or reject its findings, and whether agent-assisted changes lead to different rates of rework or incidents. Those results are not supplied in the report. GitHub’s figures show a rapid change in how pull requests begin; they do not yet settle how that change affects software quality or total development costs.
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 person may revise the draft before it is submitted.
How many pull requests did agents start in September?
GitHub reported 110 million agent-started pull requests in September 2026, representing more than 40% of pull requests on the platform that month, according to its report.
Does a high merge rate prove agent-written code is good?
No. GitHub says 90% of detected agent pull requests were merged within two weeks, but merging is not a measure of correctness or long-term quality. The report does not provide comparative defect or rework rates.
How much does GitHub estimate AI code review costs?
GitHub estimates an average Copilot review at $0.25 for Lite, $1.50 for Balanced and $10 for Max. GitHub Actions minutes are additional.
Who is responsible for code 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 request, and the person who requested the change cannot approve it.
Source: ThorstenMeyerAI.com
