Chapter 2: The Evolution of Product Teams

You lead a product organisation at a scaling technology company. Your teams follow the standard industry structure of pods containing one product manager, one designer, and six engineers. You operate on quarterly planning cycles. You spend thirty percent of your time in alignment meetings. You wait weeks for data analysts to answer basic queries. You watch engineers wait for design mocks. You observe product managers spending their days writing requirements that no one reads in full. This structure worked during the period of low interest rates and abundant capital. It does not work now. The pace of technological change creates a crisis of speed. Your competitors use artificial intelligence to ship new capabilities weekly. They operate with half your headcount. They do not have layers of middle management. They have collapsed the silos between functions. You feel the weight of coordination tax slowing every decision. You realise that your org chart creates friction rather than value. You face a fundamental shift in how humans and machines collaborate to build software. You must decide whether to maintain your traditional hierarchy or dismantle it to survive. The tension lies between the safety of proven structures and the efficiency of a new era. You need to transition from a project-based factory to a dynamic network of high-agency builders. If you do not evolve your team structure now, your product will become a relic of a slower age.

CORE SKILL OR PRINCIPLE

The evolution of product teams requires a shift from managing people to orchestrating task-based networks. You must stop viewing your product as a static artifact and start viewing it as a living organism. A traditional product team ships a feature and moves to the next project. An evolved team builds systems that learn and learn from every user interaction. This requires a transition to the full stack builder model. In this model, you dissolve the boundaries between product management, design, and engineering. You empower individuals to take an idea from insight to launch regardless of their functional title. You replace rigid org charts with fluid pods that reassemble based on the mission. You differentiate between fast thinking and slow thinking work streams. Fast thinking teams ship new capabilities weekly to find product market fit for novel features. Slow thinking teams focus on durable growth and system reliability. Success in this era demands that CEOs return to being individual contributors. You must get into the weeds of the technology to understand what your product should become. You empower local CEOs who own their parts of the business end to end. You build teams based on the Beatles principle of four to six people to maximise scenius. You create career paths for super individual contributors who lead through craft rather than management. This approach reduces coordination costs and increases the metabolism of your organisation.

EVIDENCE FROM THE CONVERSATION

The shift from product as artifact to product as organism defines the current technological transition. Modern products no longer function as static sets of features shipped into the world. They ingest data, tune themselves to outcomes, and improve through every interaction. This creates a new form of intellectual property where products think, live, and learn. As software primitives change, the model sits alongside the code as a core component. This evolution means the user interface can shift and evolve automatically based on usage patterns.

Successful companies embrace AI fluency across the entire organisation. They use AI to raise the ceiling for top talent and lower the floor for complex tasks. In these organisations, the org chart begins to resemble the work chart. Hierarchy matters less than task-based opportunities. Systems become an ensemble of models and humans grouped in different configurations to solve problems. This leads to a flattening of organisations as the marginal cost of good output approaches zero.

The full stack builder model empowers individuals to develop experiences end to end. It focuses on fluid interaction between human and machine rather than a sequence of manual handoffs. At LinkedIn, this meant scrapping the traditional APM program for an associate product builder program that teaches code, design, and product management simultaneously. This model allows anyone to take an idea to market regardless of their functional role.

Airtable implemented a reorganisation into fast thinking and slow thinking groups. The fast thinking group ships new capabilities on a near-weekly basis to provide immediate user value. The slow thinking group focuses on turning those seeds of adoption into durable growth. This structure acknowledges that AI evolves so rapidly that it implies novel form factors must be invented constantly.

CEOs are becoming individual contributors again to navigate this paradigm shift. They spend time in the code and lead specific technical initiatives themselves. This hands-on approach allows leaders to rebuild their intuitions. You cannot delegate the understanding of model capabilities or technical constraints. Brian Chesky notes that the CEO should effectively function as the chief product officer.

Revolut uses a local CEO model where product owners have end-to-end responsibility for their part of the business. These leaders define roadmaps, contribute to strategy, and execute relentlessly. Engineers and designers report directly to them as line managers. This creates a flat hierarchy where founders remain hands-on in the details of every product review.

Small teams outperform large committees. The Beatles principle suggests that four people create a collective idea of genius, or scenius, that twenty-four people cannot replicate. The original Macintosh team consisted of only twenty people. Small, senior teams move faster because they reduce communication overhead and Dunbar’s number effects. Thermal teams provide a similar benefit by isolating small groups from corporate bureaucracy to prove out zero-to-one ideas.

The rise of the super individual contributor offers an alternative to the management track. These roles, sometimes called captains, lead from the field rather than the sidelines. They tackle the most strategic problems while managers focus on coaching and positioning. This prevents a management death cycle where talented builders feel forced into people management to progress.

PRACTICAL BREAKDOWN

You must restructure your teams to eliminate the bottleneck of functional silos. Start by identifying your high-agency builders who can flex across disciplines. These are your full stack builders. You must provide them with a platform that allows for rapid prototyping and deployment.

Organise your teams into pods of four to six people. This size maintains high velocity and clear communication. Ensure each pod has a clear mission and a single decision-maker. This leader acts as the local CEO. They must own the business outcome, not just the delivery of features. You must hold them accountable to results rather than their methodology.

Adopt the fast thinking and slow thinking team structure. Assign your most creative and technical talent to the fast thinking group. Task them with shipping one new AI-powered capability every week. Do not require long PRDs or extensive roadmap planning for this group. Use their output to learn what users actually want. Once a feature shows retention, move it to the slow thinking group for scaling and integration.

As a leader, you must transition to the IC CEO model. Block out time every week to play with every new AI product in your category. Write prompts. Build prototypes. Do not delegate your understanding of the technology. You must set the pace for the organisation by making faster decisions. Use your reviews to teach your team how to think through problems from first principles.

Create a career track for super ICs. Match their compensation to that of senior managers. Give them the hardest strategic problems to solve. This keeps your best crafters on the tools. It prevents you from losing star players to the administrative overhead of management.

Use agents to turn your org chart into a task network. Identify repetitive workflows such as meeting summaries, data analysis, or Jira updates. Assign these tasks to AI agents managed by your builders. This collapses the stack and allows a single person to do the work that previously required a department.

PRACTICAL APPLICATION

Apply the local CEO model to your next product launch. Give one person the authority to make all trade-off decisions between engineering, design, and marketing. Remove the requirement for them to get consensus from twenty stakeholders. This person must understand the customer deeply and the business model thoroughly. They should be able to stand in a meeting with a major customer and make commitments on the spot.

Implement the Beatles principle during your next zero-to-one project. Select four individuals with complementary spikes in their skill sets. One might spike in technical architecture, another in visual craft, another in market strategy. Do not add more people as the project gains complexity. Force the team to use AI tools to handle the increased load. This maintains the team's chemistry and prevents the dilution of ownership.

Execute a fast thinking cycle for a new AI feature. Set a deadline of one week. Require the team to show a working prototype rather than a slide deck. Use this to bypass the translation phase of product development. When the prototype is ready, put it in front of ten lighthouse users immediately. Use their qualitative feedback to decide whether to kill the idea or double down.

Manage your leadership state by setting personal SLAs. Ensure you are never the bottleneck for a team decision. If a team needs a sign-off to move forward, provide it within hours, not days. This high clock speed trickles down through the organisation and sets a standard for urgency.

Transition your management style to coaching. Spend your one-on-ones helping your team improve their craft rather than checking on their status. Use a framework like the GROW model to ask powerful questions about their goals and options. This builds the capacity of your team to make great decisions without you.

ACTION CHECKLIST

- Evaluate your current team size and identify any group larger than eight people.

- Select three high-agency builders to pilot the associate full stack builder program.

- Block four hours on your calendar this week for hands-on prototyping with new AI tools.

- Define the single objective for your next algorithm or AI feature and write it on a shared board.

- Identify one "fast thinking" project that can be shipped as a prototype within seven days.

- Assign a single-threaded owner to your top three most critical user journeys.

- Audit your meeting schedule and cancel any recurring meeting that does not drive next steps.

- Create a "captain" role in your engineering or product ladder with compensation equal to a director.

- Conduct a "walk the store" review where you experience your product exactly as a new user does.

- Identify five "paper cuts" or small frictions in your developer experience and assign a pod to fix them this week.

- Set a personal SLA to respond to all team blockers within four hours.

- Move the desks of your product trio so they sit together with physical or virtual walls for deep work.

- Write a one-page biannual strategy that replaces your complex quarterly OKR planning.

- Identify one task that can be fully automated by an AI agent and assign a builder to implement it.

- Recruit ten "lighthouse users" for your next experimental feature and create a direct Slack channel with them.

- Ask your managers if they would rather have one more headcount or an elite AI coding assistant subscription for the whole team.

- Switch your product review format from presentations to two-way collaborative write-ups.

- Define who you are NOT solving for in your current product strategy to create a clear guardrail for your team.

- Implement a "tag-up" ritual where triads meet with their braintrust for thirty minutes weekly to unblock decisions.

- Map your core product functions against what AI can currently automate to identify augmentation opportunities.

- Commit to a team structure where five teams do one thing rather than one team doing five things.