Why LinkedIn is turning PMs into AI-powered "full stack builders” | Tomer Cohen (LinkedIn CPO)
Tomer Cohen, LinkedIn's CPO, discusses their 'Full Stack Builder' model, a new approach to product development that leverages AI to empower anyone to take ideas from concept to launch. This aims to increase agility and adapt to the accelerating pace of technological change.
Deep Dive Analysis
14 Topic Outline
The Necessity of Rethinking Product Development
The Full Stack Builder Model: Vision and Core Principles
Key Traits for Builders in an AI-Powered World
Three Pillars of Full Stack Builder Implementation: Platform, Tools, and Culture
Re-architecting Platforms for AI Integration
Developing Specialized AI Agents for Product Development
Customizing Off-the-Shelf AI Tools for Enterprise Use
Pilot Program Results and Early Adoption Trends
Strategies for Effective Change Management and AI Adoption
Evolving Performance Reviews for Full Stack Builders
Addressing Challenges and the Role of Specialization
Cultivating Full Stack Builder Talent and Mindset
Advice for Implementing AI-Driven Product Transformation
Tomer Cohen's Reflections on His Time at LinkedIn and Future Plans
7 Key Concepts
Full Stack Builder Model
An approach to product development that empowers individuals to take an idea from conception to launch, combining traditional PM, design, and engineering skills, often with the fluid interaction of human and AI. This model aims to collapse the complex, specialized product development stack back into a more nimble, adaptive, and resilient process.
Builder Key Traits
The essential human skills that great builders should focus on in an AI-augmented world: vision (compelling future stance), empathy (understanding unmet needs), communication (aligning others), creativity (beyond the obvious), and judgment (high-quality decisions in ambiguity). Everything else is intended for automation.
AI Product Development Pillars
The three fundamental components required to implement the Full Stack Builder model at scale: a re-architected 'Platform' for AI reasoning, specialized 'Tools and Agents' to automate tasks, and a supportive 'Culture' to drive adoption and mindset shift.
Trust Agent
An internal AI tool developed at LinkedIn that analyzes product specifications and ideas to identify potential vulnerabilities, harm vectors, and trust-related issues unique to the platform. It helps proactively prevent problems that might otherwise be missed until later stages.
Growth Agent
An internal AI tool at LinkedIn trained on the company's unique growth loops, funnels, and historical test data. It helps teams critique ideas for growth potential and can be used by user research teams to identify high-impact opportunities.
Research Agent
An internal AI tool at LinkedIn trained on member personas, past research, and support tickets. It helps teams understand specific user segments (e.g., small business owners, job seekers) and provides critiques on product ideas from their perspective.
Navy SEALs Analogy for Teams
A mental model for future organizational structure, where small, cross-trained pods of individuals specialize in a specific mission. This structure emphasizes nimbleness, quick assembly, and adaptability, contrasting with large, functionally siloed teams.
7 Questions Answered
LinkedIn is rethinking its process because the pace of change in job skills and market demands is accelerating, making traditional, complex, and specialized product development models too slow and unadaptive to remain competitive.
The program aims to empower individual builders to take an idea from conception to launch, leveraging AI and automation to handle many traditional sub-steps, allowing humans to focus on vision, empathy, creativity, communication, and judgment.
LinkedIn finds that off-the-shelf AI tools (like Copilot or Figma Make) never work effectively on their enterprise code or design systems without significant internal customization and re-architecting of their core platforms to allow AI to reason over them.
Contrary to expectations that AI would primarily uplift lower performers, LinkedIn's pilot shows that top talent tends to adopt AI tools fastest, driven by an innate desire to continuously improve their craft and stay at the cutting edge.
LinkedIn is incorporating 'AI agency and fluency' into performance evaluations, signaling that using and improving AI tools is a valued part of an employee's contribution and career growth.
The APB program is LinkedIn's replacement for its traditional Associate Product Manager (APM) program, designed to teach new hires coding, design, and product management skills together, preparing them to be full stack builders from the start.
No, Tomer Cohen states that while the Full Stack Builder mindset is encouraged, specialization still has a place and role within the organization, though fewer specialized roles may be needed than in the past.
26 Actionable Insights
1. Reimagine Building First Principles
To stay competitive, companies must revisit first principles and reimagine their approach to building products, as the pace of change is faster than the ability to respond.
2. Prioritize AI Change Management
Actively manage change by building incentive programs, providing motivation, and showcasing successful examples to foster a culture where people see the value and want to adopt AI tools.
3. Invest in Platform, Tools, Culture
Successfully implementing AI-driven product development requires investment in re-architecting core platforms for AI, building custom AI tools/agents, and fostering a supportive culture for adoption.
4. Empower End-to-End Builders
Empower builders to develop experiences from idea to launch, combining skills across traditional domains, rather than segmenting the process into complex sub-steps and specialized roles.
5. Focus on Core Human Traits
Builders should prioritize developing vision, empathy, communication, creativity, and especially judgment, as these human traits are crucial and hard to automate.
6. Automate Non-Core Development Tasks
Actively work to automate all product development tasks that do not require core human traits, freeing up builders for higher-value, strategic work.
7. Foster Human-AI Fluid Interaction
Shift from sequential product development to a fluid interaction between human builders and AI tools, integrating AI at every step for efficiency and quality.
8. Re-architect Platforms for AI
Re-architect core platforms and codebases (e.g., building composable UI components) to enable AI to reason over them effectively, as off-the-shelf AI tools often require significant customization.
9. Customize Third-Party AI Tools
Do not expect third-party AI tools to work out-of-the-box; plan to customize them extensively and collaborate with vendors for effective integration with your unique tech stack.
10. Build Custom AI Agents
Develop specialized AI agents (e.g., trust or growth agents) trained on your company’s unique data, context, and know-how, as generic agents cannot capture specific organizational needs.
11. Curate AI Training Data
Carefully curate specific ‘gold examples’ and define the context window for AI training, rather than giving access to all internal data, to avoid hallucinations and misinterpretations.
12. Automate Code-to-Launch Processes
Invest in AI agents for the ‘code to launch’ phase, including coding, maintenance (for failed builds), and QA agents, to significantly accelerate development and reduce manual effort.
13. Adopt Small, Nimble Pods
Organize into small, cross-functional ‘pods’ of full-stack builders who can flex across different roles, focusing on specific missions for a quarter to increase velocity and nimbleness.
14. Adjust Performance Reviews for AI
Align performance expectations, hiring criteria, and evaluation processes to reward ‘AI agency and fluency,’ encouraging employees to actively use and improve AI tools.
15. Pilot and Showcase Successes
Implement pilot programs to demonstrate tangible successes of AI-driven workflows, providing concrete examples that inspire broader adoption within the organization.
16. Develop Full-Stack Training Programs
Institute formal training programs (e.g., Associate Product Builder program) to teach new hires how to code, design, and product manage, fostering a new generation of full-stack builders.
17. Act Without Formal Directives
Don’t wait for a formal reorg or declaration; proactively use or build new tools, demonstrate a full-stack builder mindset, and prove capabilities through action.
18. Maintain High Visibility During Pilots
Even when starting with a small core team for AI initiatives, ensure high visibility across the organization by regularly sharing progress, tools, and early successes to foster awareness and engagement.
19. Be Patient with AI Transformation
When transforming a large organization with AI, be ambitious about the goal but patient and thoughtful about implementation, understanding it requires significant upfront investment and time.
20. Allocate Resources for AI Infrastructure
Be fully aware of the necessary investment in re-architecting platforms and customizing AI tools, and ensure adequate resources are allocated to these foundational efforts for successful outcomes.
21. Declare AI Vision, Continuous Progress
Clearly communicate the long-term vision for AI adoption and emphasize that it’s a continuous journey of improvement, not a fixed end state, to align the organization and foster a growth mindset.
22. Over-communicate AI Progress
Over-communicate not only the vision for AI transformation but also the ongoing progress, potentially using KPIs or OKRs, to maintain transparency and hold leadership accountable.
23. Proactively Engage with AI
Individuals should proactively engage with AI tools and new building methods within their current role or seek opportunities in organizations that are at the cutting edge of AI-driven development.
24. Continuously Improve Your Craft
Top talent consistently strives to enhance their skills and stay at the cutting edge of their craft, which is crucial for adapting to rapid changes in job requirements.
25. Highlight Cross-Functional Transitions
Showcase examples of employees leveraging AI tools to transition into new roles or functions, demonstrating career growth opportunities enabled by full-stack building.
26. Leverage AI for Empowerment
View technology, especially AI, as a tool for empowerment and meritocracy, enabling individuals to achieve more and contribute based on their abilities rather than their specialized role.
7 Key Quotes
When we look at the skills required to do your job, by 2030, they will change by 70%. So whether or not you're looking to change your job, your job is changing.
Tomer Cohen
The work itself is not complex, but the process we made very complex.
Tomer Cohen
It's not enough to give them the tools. You have to build the incentives programs, the motivation, the examples to how you do it.
Tomer Cohen
Top talent has this tendency of continuously trying to get better at their craft.
Tomer Cohen
It never works. Never works. You have to bring it in and customize a lot of it, working almost in alpha mode with those companies to make it work internally.
Tomer Cohen
If you're looking for a formal reorg or declaration to start building differently, you're waiting too long.
Tomer Cohen
Becoming is better than being.
Tomer Cohen
1 Protocols
Full Stack Builder Implementation
Tomer Cohen- Invest in Platform: Re-architect core platforms (e.g., composable UI components, server-side) so AI can reason over them, as third-party tools won't work off-the-shelf.
- Develop Tools and Agents: Build or heavily customize specialized AI agents (e.g., Trust Agent, Growth Agent, Research Agent, Analyst Agent) that incorporate unique company knowledge and context.
- Cultivate Culture: Implement change management strategies, including setting new performance expectations (AI agency/fluency), piloting and celebrating internal successes, creating exclusive programs (APB), and fostering an environment where people are encouraged to explore and share AI tools.