Chapter 32: The Path Forward

You lead a product organisation at a software firm that has dominated its category for a decade. Your roadmap for the next eighteen months is detailed and approved by your board. Your teams focus on user interface polish and legacy database integrations. Suddenly a frontier artificial intelligence lab releases a new reasoning model. This model automates eighty per cent of the manual tasks your team planned to build over the next year. Your competitors adopt this model in one week. Your product suddenly looks like an artifact of a previous era. You realise that your technical foundation is not a stable ground but a vertical journey. You face a decision point that defines your career and the survival of your firm. You can continue to optimise for the present world or you can begin to architect for a world of cognitive abundance. The tension lies between the safety of incremental planning and the volatility of exponential shifts. You must decide if you will be a leader who reacts to updates or an architect who understands the trajectory toward transformative intelligence. If you do not prepare for the arrival of superintelligence your product will likely be automated away by a machine that performs better than your staff. You are currently at the flat part of an exponential curve that is about to turn vertical.

CORE SKILL OR PRINCIPLE

The core principle of the path forward is transformative preparedness through model maximalism. You must internalise the fact that the artificial intelligence you use today is the worst you will ever use for the rest of your life. Success in this era depends on your ability to move from a deterministic mindset to a probabilistic one. You must move from being a manager of human tasks to an architect of human agent systems. This requires you to adopt a mindset where your own growth as a leader is the primary constraint on your company. You must prioritise taste and judgment as your primary differentiators in a world of free pixels and free code. Transformative preparedness is the skill of maintaining extreme clarity while operating in a state of constant change. You must view your technical architecture as the primary driver of your strategy. True leadership involves taking responsibility for the alignment and behaviour of digital workers. You must build a culture that thrives on rapid learning cycles and aggressive prototyping.

EVIDENCE FROM THE CONVERSATION

Evidence from leaders at OpenAI and Anthropic shows that model capabilities are evolving on a two month cadence. Kevin Weil observes that computers can now do things every sixty days that were previously impossible. This pace renders traditional eighteen month roadmaps obsolete. Amjad Masad at Replit reports that his team slaughtered their entire roadmap when new computer use capabilities dropped. He argues that founders must be resilient to change and ready to switch priorities immediately. The concept of model maximalism is supported by the fact that scaling laws are still accelerating. Benjamin Mann predicts a fifty per cent chance of reaching superintelligence by 2028. He measures this progress through the economic Turing test which determines if a machine can perform a three month job without the employer noticing. When fifty per cent of money weighted jobs pass this test the world enters the era of transformative intelligence.

Evaluation of company potential requires looking at the behaviour of leadership. Dan Shipper finds that if a chief executive uses tools like ChatGPT or Claude daily the organisation succeeds. Tomer Cohen at LinkedIn notes that top talent uses these tools to get better at their craft while lower performers may use them to go on autopilot. He has replaced the traditional associate product manager program with an associate full stack builder program to lean into this shift. The counterintuitive truth about best practices is that rote building is now a commodity. Marily Nika warns against the shiny object trap of doing artificial intelligence for the sake of excitement. Aishwarya Reganti finds that seventy five per cent of enterprises struggle with reliability when they follow hype instead of obsessing over customer problems.

Finally evidence suggests that following efficiency best practices can accelerate labour disruption. Kunal Shah observes that inefficiency is currently the world's largest employer. If leaders perfectly automate every inefficient workflow they remove the need for existing roles. This creates a world where twenty per cent unemployment may be inevitable as machine intelligence replaces white collar labour. Marty Cagan identifies that product owners who act as backlog administrators are the most vulnerable to this shift.

PRACTICAL BREAKDOWN

WHY TODAY'S AI IS THE WORST YOU'LL EVER USE AGAIN

You must recognise that your current dissatisfaction with model performance is temporary. Models improve with more data and better post training techniques. You must build your product to get better automatically as foundational models get smarter. Do not build features that compensate for current model weaknesses because those features will become technical debt when the model improves. Focus your engineering effort on proprietary data sets that provide unique context. This ensures that your product is a gift that keeps on giving as the underlying intelligence evolves.

MODEL MAXIMALISM AND EMBRACING RAPID CHANGE

You must adopt a model maximalist strategy by consuming the latest intelligence immediately. Stop trying to create foundational models yourself unless you have a massive data advantage. Move from an imperative model where you give step by step instructions to a declarative model where you describe desired outcomes. Implement a ten per cent planning rule where you spend no more than one week planning for a ten week execution cycle. This prevents you from being trapped in a plan that technology has already surpassed. You must be comfortable with the messy middle of innovation where things are hectically unstable.

EVALUATING COMPANY POTENTIAL

You must audit your organisation to see if it is maximising its potential. Check if your chief executive uses reasoning models multiple times per hour. Verify if your performance reviews reward people for using artificial intelligence to increase their span of control. Look for teams that have replaced static documents with working prototypes. If your teams are still waiting for data scientists to run basic analysis you are failing to empower your staff. Success is predicted by how much high agency curiosity exists at the edge of your company.

THE COUNTERINTUITIVE TRUTH ABOUT BEST PRACTICES

You must ignore the marketing hype of one click agents and generic chatbots. True success comes from deconstructing and reconstructing your core workflows from first principles. You must be obsessed with the problem first and the technology second. Recognise that implementation is becoming ridiculously cheap which makes design and taste more valuable. You must appoint a benevolent dictator with great taste to be the ultimate judge of model performance. This individual ensures that your agents do not produce AI slop that chases dopamine instead of truth.

THE UNEMPLOYMENT PARADOX

You must understand that high performance leads to job displacement in a world of abundant intelligence. If you remove all friction and inefficiency you remove the need for people who manage that friction. Your role shifts from managing people to managing model behaviour and intent. You must hire for high agency and curiosity rather than specific technical skills that will be automated. Tell your new hires that their job is to manage a team of junior digital workers.

SKILL APPLICATION

Apply these path forward skills by running a pre mortem for your product in a world of cognitive abundance. Assume that code and pixels are free and identify what unique human value remains. This value is usually your brand trust your relationships and your proprietary data flywheels.

Operationalise model maximalism by requiring every team to show how they used a new capability in their weekly demo. If a new model release makes a feature obsolete kill that feature immediately and pivot the team. Do not allow your past investments to dictate your future strategy.

Implement the associate full stack builder model for your next project. Select a small team and give them the resources to take an idea from insight to launch independently. Encourage your product managers to write code and your engineers to design interfaces. This collapses the silos and increases metabolic speed.

Manage your leadership presence by becoming an IC executive. Block out time every day to build functional prototypes yourself. This hands on work prevents you from losing your technical intuition. It also sets an example for your team that curiosity and play are professional requirements.

Foster an agentic society within your company by delegating repetitive tasks to digital workers. Use these agents to summarise meetings and route tickets and draft strategy memos. This frees your human staff to focus on high level systems thinking and creative problem solving.

AVOIDING THE LOW IMPACT PM DEATH SPIRAL

You must identify if you are caught in the low impact product management death spiral. This occurs when your primary contribution is administering a backlog in Jira rather than solving user problems. You must move from being a process person to being a taste maker and editor. Start by documenting the rationale behind every decision you make in a decision log. Evaluate whether your work moves value and viability metrics or just ships features. If you find yourself in a feature team you must actively raise your skills or find an environment that values empowerment. You must solve the right problem rather than just building the right product.

BUILDING FOR THE NEXT DECADE OF PRODUCT MANAGEMENT

You must prepare for the next decade where AI will be as common as the database. Every product manager will effectively become an AI product manager. This requires you to get comfortable with the uncertainty of research and the non determinism of models. You must build your unique talent stack around what is uniquely human. Focus on creative thinking and empathy and influence. Your value will come from your ability to connect dots across diverse fields to develop new paradigms of interaction. The bottleneck of the future is the generation of ideas rather than the making of things. You must become a generative leader who can articulate a vision and align a human agent medley to achieve it.

ACTION CHECKLIST

• Calculate the money weighted percentage of your company's tasks that could pass the economic Turing test today.
• Schedule a meeting with your chief executive to verify if they use reasoning models multiple times per day.
• Identify one project that technology has made obsolete and kill it this week.
• Write a one page constitution for your team's usage of digital workers.
• Block out two hours this week to build a functional prototype using a vibe coding tool.
• List the three things your product is not for to create a clear guardrail for your team.
• Replace your next recurring status meeting with an automated summary generated by an agent.
• Audit your hiring plans and remove any role that an agent will likely perform in twenty four months.
• Ask your lead engineer to identify one piece of strategic technical debt you can leverage for speed.
• Set a personal service level agreement to respond to all team blockers within four hours.
• Define what quality means for your product using a specific rubric and share it today.
• Create a Slack channel for sharing provocative generative experiments found in the community.
• Record a customer interview and share a five minute highlight reel with your builders.
• Set an arbitrary deadline trap for an upcoming feature to force scope prioritisation.
• Ask yourself if you would fully fund your own team if you were the chief executive today.
• Identify one task currently done by a human that a digital worker could automate this week.
• Commit to a six week execution cycle for your next major initiative and use a circuit breaker.
• Write your personal operating manual or README and share it with your team.
• Conduct a walk the store review of your primary user journey as a new user.
• Use the what so what now what structure for every update you send this week.
• Commit to the nail it then scale it philosophy by refusing to add headcount until a product has found level four fit.