Personal Productivity & AI-Augmented WorkLenny's Newsletter
How Intercom 2x’d their engineering velocity in 9 months with Claude Code | Brian Scanlan
ai-coding-toolscursor-vs-copilotautomation-stacks
“100% of engineers—plus designers, PMs, and TPMs—now shipping code via Claude Code”
Key takeaways
- Intercom doubled engineering throughput (merged PRs per R&D employee) in 9 months using Claude Code while maintaining code quality
- Built custom telemetry infrastructure to measure AI adoption and quality impact across hundreds of engineers, plus skills repository with automated enforcement hooks
- Achieved 100% adoption across engineering AND expanded to non-technical roles (designers, PMs, TPMs) shipping code—suggesting AI coding tools democratize development
- Preparing for agent-first world with CLIs, MCPs (Model Context Protocol), and ephemeral APIs—architectural shift beyond just productivity gains
- Permission and accountability framework enabled rapid adoption; 'backlog zero' now achievable, fundamentally changing engineering culture and planning
Why this matters for operators: Engineering leaders evaluating AI coding tool ROI and adoption strategies; companies considering Claude vs Cursor/Copilot
I cover AI×GTM intelligence like this every Wednesday.
Get STEEPWORKS WeeklyMore picks
AI EcosystemsSemafor
California city bans data center construction as opposition grows nationwide
- Monterey Park, CA became first US city to permanently ban data center construction with 86% voter support
- Public opposition to nearby data centers nearly doubled from 42% to 71% in just nine months
- Growing anti-AI sentiment is strongest among young people experiencing AI-driven labor market impacts
regulatory-impactai-policyai-infrastructure-backlash
AI×GTMGTM AI Podcast & NewsletterVictor's pick
6/4/26: Why and How to run AI with NO Internet
Where we maybe previously paid the W-2 of a human to do this necessary thing for the business, that cost didn’t really go away. It just transferred from a W-2 to an inference provider.”
- GTM operators are entering a 'toolbox era' where bringing your own AI stack (like mechanics bring tools) becomes expected in FTE and fractional roles
- Running AI models locally (Ollama, LM Studio, Jan.ai) gives operators data ownership and independence from SaaS vendor terms of service and uptime
- GitHub repos are becoming the new resume for GTM operators - demonstrating technical capability and owned infrastructure matters more than traditional credentials
local-ai-deploymentdata-sovereigntyoperator-toolbox-era
GTM OpsGTM AI Podcast & NewsletterVictor's pick
Who Owns the System that Compounds?
Great framing for functional emergence of AI in GTM
- RevOps faces existential choice: evolve into GTM system architect or be automated away by AI - no middle path exists
- Modern GTM is now a system with 106+ SaaS tools creating quadratic complexity (106 integration points per new tool) that humans cannot manually operate
- AI deployment sequence matters critically - most companies implement backward by automating tactical work before fixing underlying system architecture
revenue-platform-consolidationsignal-infrastructureback-to-basics-gtm
This analysis was produced using the STEEPWORKS system — the same agents, skills, and knowledge architecture available in the GrowthOS package.