Database › Intercom
Intercom
Intercom publicly committed to doubling R&D productivity within 12 months. It set targets for the share of pull requests driven by agents (Claude Code) and approved by AI.[1][2]
Tools
Metrics tracked
- Intercom tracks PRs per R&D person per month, the product defect backlog, idea-to-ship speed, downtime from breaking changes, cost per PR, the share of agent-driven PRs and the share of AI-approved PRs.[1][2]
Guardrails
- Intercom says it has built shared Claude Code plugins, skills and hooks to encode its standards.[3]
Hiring, reviews & process changes
- Code review changed: 19% of merged PRs are now auto-approved by AI with no human review.[2]
Reported results
- PRs per R&D person rose from 16 to 33 per month, and the product defect backlog fell 54%.[2]
- Product changes ship 39% faster, and downtime from breaking changes fell 35% even as deployments doubled.[2]
- Fully loaded cost per PR was cut in half, and 93.6% of PRs are agent-driven via Claude Code.[2]
- 1,100 of Intercom's 1,305 employees use Claude Code.[2]
Metric tags: PR throughput, Quality / defects, Adoption rate, Cost per PR
Editor's note: The mandate is classed as 'yes' because leadership set explicit, organization-wide adoption targets.
Sources
- Ideas by Fin/Intercom: 2x, nine months later · 2026-04-16 · company source
- Darragh Curran (Intercom CTO) on LinkedIn: Nine months ago, we publicly committed to 2x... · 2026-04-16 · primary source
- Ideas by Fin/Intercom: How we use Claude Code today at Intercom · 2026-03-19 · company source
Last verified 2026-10-25. Industry and size are editorial bands; see methodology. Spot an error? Submit a correction.