🎙️ This week on How I AI: How Intercom 2x’d their engineering velocity with Claude Code

2026-04-20 · Source: Lenny's Newsletter · Field: Technology & Digital — Software Development & Engineering, Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems · Depth: Advanced, extended

Summary

Intercom doubled its engineering throughput in nine months by extensively integrating Claude Code, as detailed by Senior Principal Engineer Brian Scanlan. The company achieved this by treating its engineering organization like a product, instrumenting everything from skill invocations in Honeycomb to anonymized Claude Code sessions in S3, and building custom dashboards for performance insights. This velocity gain, measured by merged PRs per R&D employee, was built upon a mature foundation of CI/CD, comprehensive test coverage, and a high-trust culture. Intercom also implemented custom skills with hooks, such as a "Create PR" skill that enforces context-rich pull request descriptions, and developed self-improving AI agents to fix flaky tests. This approach has led to improved code quality and increased capacity for addressing technical debt, with the company now aiming for a 10x increase in throughput.

Key takeaway

For Directors of AI/ML and CTOs aiming to significantly boost engineering velocity, you should prioritize establishing a robust CI/CD pipeline and comprehensive test coverage before scaling AI tool adoption. Instrument your AI workflows with detailed telemetry to identify bottlenecks and opportunities for custom skill development. Empower your teams with explicit permission to experiment with AI, fostering a culture where accountability rolls up, enabling rapid iteration and a focus on agent-first work to tackle technical debt and accelerate product delivery.

Key insights

Intercom doubled engineering velocity in nine months by deeply integrating Claude Code, emphasizing telemetry and cultural permission.

Principles

• Treat your engineering organization like a product.
• AI magnifies existing strengths and weaknesses.
• Technical leadership must grant permission and take accountability.

Method

Instrument AI tool usage with telemetry (Honeycomb, S3) for visibility. Implement custom AI skills with hooks to enforce quality and automate workflows. Design agent-friendly product interfaces.

In practice

• Track skill invocations and session data for AI adoption insights.
• Build guardrails with custom skills for quality enforcement.
• Prioritize fixing foundational issues before scaling AI adoption.

Topics

• Claude Code
• Engineering Velocity
• AI Adoption Strategy
• Custom AI Skills
• Telemetry & Analytics

Best for: Software Engineer, Director of AI/ML, CTO

Related on AIssential

• How Intercom 2X'd engineering velocity with Claude Code | Brian Scanlan — Intercom doubled R&D throughput and improved code quality by fully integrating AI-assisted development and treating internal AI adoption as a product.
• How Intercom 2x’d their engineering velocity in 9 months with Claude Code | Brian Scanlan — Deep AI integration and telemetry can double engineering throughput and improve code quality.
• v2.1.97 — Claude Code prioritizes security, performance, and extensibility through continuous iterative improvements and new feature introductions.
• We're sharing our internal AI engineering cheatsheets — Pre-made engineering decisions and architectural guidelines accelerate AI development and improve output quality.
• How a non-technical project manager built and shipped a stress management app with Claude Code in six weeks — Non-technical users can rapidly build and ship complex applications using AI code assistants like Claude Code.
• DXC will integrate Claude into the systems banks, airlines, and other regulated industries rely on — Regulated industries can safely integrate advanced AI like Claude through specialized engineering and proven internal adoption.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Lenny's Newsletter.