<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/6gKsU5Fpz" -->

---
title: How LinkedIn Uses AI to Review 40,000 Pull Requests a Week
description: LinkedIn built an in-house AI code reviewer that covers more than 40,000 pull requests a week across its 10,000-plus repositories. Most reviews post before a...
canonical: https://daily.dev/posts/how-linkedin-uses-ai-to-review-40-000-pull-requests-a-week-6gksu5fpz
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: How LinkedIn Uses AI to Review 40,000 Pull Requests a Week | daily.dev
og:description: LinkedIn built an in-house AI code reviewer that covers more than 40,000 pull requests a week across its 10,000-plus repositories. Most reviews post before a...
og:url: https://daily.dev/posts/how-linkedin-uses-ai-to-review-40-000-pull-requests-a-week-6gksu5fpz
og:image: https://api.daily.dev/og/posts/6gKsU5Fpz.png
og:image:alt: How LinkedIn Uses AI to Review 40,000 Pull Requests a Week
og:image:width: 1200
og:image:height: 630
og:locale: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# How LinkedIn Uses AI to Review 40,000 Pull Requests a Week

**[ByteByteGo](https://daily.dev/sources/bytebytego)** · 16 min read · 13 upvotes · 4 comments

## Summary

LinkedIn built an in-house AI code reviewer that covers more than 40,000 pull requests a week across its 10,000-plus repositories. Most reviews post before a human even opens the PR. The system uses a multi-agent orchestrator-subagent design, with independent subagents reviewing the same diff, an orchestrator merging overlapping findings, and a separate model verifying single-subagent findings to catch hallucinations. Rules are layered into company-wide, repository, and PR-specific tiers. In production, engineers accepted 63.9% of AI comments, with 100% acceptance on concurrency bugs and dead code. LinkedIn is now building a risk heat map to flag the riskiest diff regions first.

## Full article

daily.dev links to this article rather than hosting it. Read it at the original source: <https://blog.bytebytego.com/p/how-linkedin-uses-ai-to-review-40000>

## Questions this post answers

### How many pull requests does LinkedIn's AI code reviewer handle per week?

LinkedIn's in-house AI code reviewer covers more than 40,000 pull requests in a typical week across its more than 10,000 active code repositories. Most of its reviews are posted before the first human reviewer even opens the PR, removing review as the main bottleneck after coding agents sped up writing code.

_Teams scaling code review automation can track engineering writeups like this on daily.dev._

### What is the orchestrator-subagent pattern for multi-agent AI code review?

It's a design where one orchestrator agent dispatches a PR diff to several subagents that review independently without seeing each other's findings, each returning structured findings with line numbers. The orchestrator then merges overlapping findings as higher-confidence signals and sends single-subagent findings to another model for verification before posting, reducing hallucinations compared to using one model alone.

_Developers designing multi-agent review pipelines can find similar architecture breakdowns on daily.dev._

### What acceptance rate did LinkedIn achieve for its AI code review comments?

Engineers accepted 63.9% of thousands of AI-generated review comments measured in production, where acceptance means the code actually changed because of the comment, not just a thumbs-up reaction. Acceptance was highest for concurrency bugs and dead code at 100% and logic errors at 80%, metrics LinkedIn tracks daily to decide which rules and prompts to keep or change.

_Anyone benchmarking AI reviewer quality can follow real acceptance-rate data like this on daily.dev._

## Community discussion

Top comments from developers on daily.dev.

**@yildirim** · 3 upvotes

> 40k PR per week? What are they developing exactly?

## Similar posts on daily.dev

- [AI Code Review at Scale: LinkedIn's Multi-Agent Approach](https://daily.dev/posts/ai-code-review-at-scale-linkedin-s-multi-agent-approach-7lpkuhkzi) · InfoQ · 0 upvotes · 0 comments
- [Why AI Code Reviews Are Slowing Down PRs](https://daily.dev/posts/why-ai-code-reviews-are-slowing-down-prs-yubjqxjqr) · Aviator Blog · 0 upvotes · 1 comments
- [AI Code Review Bottleneck 2026: Why PRs Are Stuck in Review](https://daily.dev/posts/ai-code-review-bottleneck-2026-why-prs-are-stuck-in-review-o7qcycgq0) · Alex CloudStar · 2 upvotes · 1 comments

---

Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#llm](https://daily.dev/tags/llm), [#cicd](https://daily.dev/tags/cicd), [#ai-coding](https://daily.dev/tags/ai-coding), [#linkedin](https://daily.dev/tags/linkedin)

[View this post on daily.dev](https://daily.dev/posts/how-linkedin-uses-ai-to-review-40-000-pull-requests-a-week-6gksu5fpz)

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]}
{"@context":"https://schema.org","@type":"TechArticle","headline":"How LinkedIn Uses AI to Review 40,000 Pull Requests a Week","url":"https://daily.dev/posts/how-linkedin-uses-ai-to-review-40-000-pull-requests-a-week-6gksu5fpz","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/how-linkedin-uses-ai-to-review-40-000-pull-requests-a-week-6gksu5fpz"},"datePublished":"2026-11-09T16:36:33.597Z","dateModified":"2026-11-09T16:37:05.419Z","description":"LinkedIn built an in-house AI code reviewer that covers more than 40,000 pull requests a week across its 10,000-plus repositories. Most reviews post before a...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/39fb1d5bd42e8127b3825be947115308?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/39fb1d5bd42e8127b3825be947115308?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"ByteByteGo","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"ByteByteGo","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/35be29234ee14d01a9cd049c52e12753","url":"https://daily.dev/sources/bytebytego"},"commentCount":4,"discussionUrl":"https://daily.dev/posts/how-linkedin-uses-ai-to-review-40-000-pull-requests-a-week-6gksu5fpz","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":13},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":4}],"keywords":"ai-agents,llm,cicd,ai-coding,linkedin","timeRequired":"PT16M"}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"ByteByteGo","item":"https://daily.dev/sources/bytebytego"},{"@type":"ListItem","position":3,"name":"How LinkedIn Uses AI to Review 40,000 Pull Requests a Week"}]}
{"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/how-linkedin-uses-ai-to-review-40-000-pull-requests-a-week-6gksu5fpz","comment":[{"@type":"Comment","text":"40k PR per week? What are they developing exactly?","datePublished":"2026-11-11T12:59:26.337Z","url":"https://daily.dev/posts/6gKsU5Fpz#c-zf3K6GwBs","author":{"@type":"Person","name":"Yasin İsa YILDIRIM","url":"https://daily.dev/yildirim","image":"https://media.daily.dev/image/upload/s--i098KjZN--/f_auto/v1727931829/avatars/avatar_EsTd0u1bqW4Q1cI9yRp5N"},"interactionStatistic":{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":3}}]}
{"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/how-linkedin-uses-ai-to-review-40-000-pull-requests-a-week-6gksu5fpz#faq","mainEntity":[{"@type":"Question","name":"How many pull requests does LinkedIn's AI code reviewer handle per week?","acceptedAnswer":{"@type":"Answer","text":"LinkedIn's in-house AI code reviewer covers more than 40,000 pull requests in a typical week across its more than 10,000 active code repositories. Most of its reviews are posted before the first human reviewer even opens the PR, removing review as the main bottleneck after coding agents sped up writing code. Teams scaling code review automation can track engineering writeups like this on daily.dev."}},{"@type":"Question","name":"What is the orchestrator-subagent pattern for multi-agent AI code review?","acceptedAnswer":{"@type":"Answer","text":"It's a design where one orchestrator agent dispatches a PR diff to several subagents that review independently without seeing each other's findings, each returning structured findings with line numbers. The orchestrator then merges overlapping findings as higher-confidence signals and sends single-subagent findings to another model for verification before posting, reducing hallucinations compared to using one model alone. Developers designing multi-agent review pipelines can find similar architecture breakdowns on daily.dev."}},{"@type":"Question","name":"What acceptance rate did LinkedIn achieve for its AI code review comments?","acceptedAnswer":{"@type":"Answer","text":"Engineers accepted 63.9% of thousands of AI-generated review comments measured in production, where acceptance means the code actually changed because of the comment, not just a thumbs-up reaction. Acceptance was highest for concurrency bugs and dead code at 100% and logic errors at 80%, metrics LinkedIn tracks daily to decide which rules and prompts to keep or change. Anyone benchmarking AI reviewer quality can follow real acceptance-rate data like this on daily.dev."}}]}
```

