The AI Record
Last year, our CTO Darragh Curran set an explicit goal: double the productivity of our entire R&D organization within 12 months. Because the faster we can build and ship, the faster our customers get the capabilities they need. Nine months later, we did it. The results have been significant across the board, but the number that matters most for this post: downtime from breaking code changes dropped 35%, even as our deployments doubled.

Intercom (renamed Fin in May 2026)Published by the organization itself (post), no named author

Post: AI is approving our pull requests: Here's how we made it safe, Intercom Blog, April 21, 2026, at The 2x effort.

  • 35% breaking change downtime reduction, nine months after the goal was set

Checked against the source [ Grade B ] Said in the last year Sells AI tools

Filed under

Intercom (renamed Fin in May 2026) sells AI or agent tools or services, so it has a reason to publish flattering numbers about AI. The passage stays in the archive; read it with that in mind.

How this was checked

Checked against the source by a first reader, then confirmed independently by a second reader who fetched the source without seeing the first reader's notes (Independent review agent (Claude Sonnet), blind second pass from passage text and citation only, no repository access, 2026-11-04).

Found verbatim in 'The 2x effort' section, split across two paragraphs. Inline hyperlinks sit on 'double the productivity… within 12 months' and 'we did it', but the text is unchanged. The next sentence in the source, 'Shipping faster made us safer.', is not in the passage.

How to cite this passage

Intercom (renamed Fin in May 2026), “AI is approving our pull requests: Here's how we made it safe”, Intercom Blog, 2026-04-21, at The 2x effort. https://www.intercom.com/blog/ai-is-approving-our-pull-requests-heres-how-we-made-it-safe Archived at https://the-ai-record.com/passages/intercom-double-productivity-r-d-downtime-dropped-35-2026

More from the same source

Our Agent is strict. It won’t approve large PRs. If a change is too big, too complex, or too broad in scope, it flags it and requires it to be broken down. Creating a direct, positive incentive for engineers to ship smaller, more incremental, well-scoped changes.

Intercom (renamed Fin in May 2026)Published by the organization itself (post), no named author

Post: AI is approving our pull requests: Here's how we made it safe, Intercom Blog, April 21, 2026, at Not just faster; safer.

Checked against the source [ Grade B ] Said in the last year Sells AI tools Passage page

AI-authored backend code had a revert rate of 0.53%, compared to 5.39% for human-authored. On the frontend, it was 0.22% versus 2.00%.

Intercom (renamed Fin in May 2026)Published by the organization itself (post), no named author

Post: AI is approving our pull requests: Here's how we made it safe, Intercom Blog, April 21, 2026, at The data.

  • 0.5% ai authored backend revert rate, April 2026
  • 5.4% human authored backend revert rate, April 2026
  • 0.2% ai authored frontend revert rate, April 2026
  • 2% human authored frontend revert rate, April 2026

Checked against the source [ Grade B ] Said in the last year Sells AI tools Passage page

We started with a controlled pilot of over 100 PRs through the AI approval pipeline. The results: zero reverts of AI-approved PRs, and a 6–16x improvement in time-to-approval at the 75th percentile. Since then, the system has scaled significantly. In the first four weeks of broader rollout, 497 PRs went fully autonomous, with Claude writing the code and our AI approval system reviewing, approving, and shipping to production.

Intercom (renamed Fin in May 2026)Published by the organization itself (post), no named author

Post: AI is approving our pull requests: Here's how we made it safe, Intercom Blog, April 21, 2026, at The data.

  • 0 ai approved pilot reverts, controlled pilot of over 100 PRs. No person in the loop
  • 497 fully autonomous prs, first four weeks of broader rollout. No person in the loop

Checked against the source [ Grade B ] Said in the last year Sells AI tools Passage page

Today, over 93% of our pull requests (PRs) across our two main codebases are Agent-driven. And over 19% are auto-approved with no human reviewer in the loop.

Intercom (renamed Fin in May 2026)Published by the organization itself (post), no named author

Post: AI is approving our pull requests: Here's how we made it safe, Intercom Blog, April 21, 2026, at Introduction.

  • 93% agent driven pr share, April 2026, two main codebases. Level not stated
  • 19% auto approved pr share, April 2026, two main codebases. No person in the loop

Checked against the source [ Grade B ] Said in the last year Sells AI tools Passage page

Our PR review Agent doesn’t treat code review as a single task. It decomposes it into separate sub-jobs, each handled by an independent sub-Agent. One assesses the quality of the problem description. Another checks whether the diff actually aligns with the stated intent. Another reviews for safety concerns. Another checks for logical correctness. Another reviews against best practices and known anti-patterns.

Intercom (renamed Fin in May 2026)Published by the organization itself (post), no named author

Post: AI is approving our pull requests: Here's how we made it safe, Intercom Blog, April 21, 2026, at How our PR review Agent works.

Checked against the source [ Grade B ] Said in the last year Sells AI tools Passage page

Something wrong? Report a correction. JSON: /api/v1/passages/intercom-double-productivity-r-d-downtime-dropped-35-2026.json