Why LinkedIn Is Replacing PMs with AI-Powered ‘Full-Stack Builders’ | Tomer Cohen
By: Tomer Cohen Host: Lenny Rachitsky Source: Lenny’s Newsletter / Lenny’s Podcast Type: podcast
Summary
Lenny’s Podcast with Tomer Cohen, longtime CPO of LinkedIn, pioneering the Full Stack Builder program. LinkedIn scrapped its traditional Associate Product Manager program and replaced it with an Associate Product Builder (APB) program teaching coding, design, and PM skills together from day one (starting January 2025). Created a formal “Full Stack Builder” title and career ladder enabling anyone from any function to take products from idea to launch. Core argument: product development is inherently simple (research, design, code, launch, iterate) but organizations turned each step into dozens of sub-steps requiring 10-15 teams and 6 months to ship a small feature. Three pillars for making FSB work: platform (tooling infrastructure), agents (specialized AI that critiques ideas and finds vulnerabilities), and culture (most important — 80% of AI adoption is change management, not technology). Key finding: top performers adopt AI tools fastest, contrary to expectations about AI being an equalizer. Off-the-shelf AI tools fail on enterprise code — curated “gold examples” work far better than giving AI access to everything. LinkedIn’s maintenance agent now auto-fixes ~50% of failed builds. Most important skills in AI era are the most human: vision, empathy, communication, creativity, and judgment.
Key Ideas Extracted
- Full Stack Builder as new role definition: Formal title and career ladder replacing traditional PM/eng/design silos — anyone from any function can take products from idea to launch
- APM → APB program shift: New hires learn coding, design, and PM from day one rather than specializing — fundamental talent development change at one of tech’s largest companies
- Process complexity is the enemy: Product development is simple but orgs made it complex — 6 months and 10-15 teams for a small feature; FSB model cuts through this
- Top performers adopt AI fastest: Contrary to expectations that AI would help lower performers catch up, the best talent uses tools most actively — AI amplifies existing capabilities rather than leveling the field
- 80% change management, 20% technology: Only ~5% of employees naturally adopt cutting-edge tools; the rest need performance review updates, public win celebrations, dedicated training, and leader demonstrations
- Gold examples > full data access: Letting AI access all internal documents failed — AI couldn’t determine importance; carefully curated best examples and highest-quality information work far better
- AI value formula: Experimentation volume × quality ÷ time to launch — all three factors must improve together
- Most human skills matter most: Vision, empathy, communication, creativity, and judgment — everything else can be automated
- Designers writing code, PMs building dashboards: Cross-functional capabilities emerging as tools mature — tasks that required waiting for other teams now done directly
- Don’t wait for permission: Whether leader or IC, waiting for formal programs means falling behind — start using AI tools immediately and demonstrate what’s possible
Notes
- Published Dec 4, 2025 on Lenny’s Podcast. Episode ~67 min.
- Sponsors: Vanta, Figma Make, Miro
- Tomer Cohen also has his own podcast: “Building One with Tomer Cohen”
- Cross-reference: Previous Lenny’s episode “How LinkedIn became interesting” (Tomer Cohen, also in our sources as
2024-09-08-linkedin-transformation-tomer-cohen.md) - Referenced: Cursor, Devin, Windsurf, Lovable, Figma, Microsoft Copilot
- LinkedIn’s maintenance agent auto-fixing 50% of failed builds is a concrete enterprise AI success metric
- Recommended books: Why Nations Fail, Outlive, The Beginning of Infinity
Raw Content
Re-scraped from Lenny’s Newsletter 2026-02-15. Full article content captured in Summary and Key Ideas above.