We're Wasting the Time and Talent of Product People

This is the first article in a three-part series exploring how AI is transforming product discovery and development.

To succeed as a Product Manager, you must juggle a dizzying array of tasks daily. The best PMs are jacks of all trades: analytical and detail-oriented, yet creative and imaginative. They're strategic, political animals while remaining approachable and collaborative.

As a result, Product Managers are some of the highest-functioning employees in tech companies. But there's a problem: we're wasting their time and talent.

The Coordination Tax

The PM is the nervous system of product development; they sit in central command, shepherding products from idea to development to market.

Coordinating across cross-functional teams is crucial. It's also the most expensive and lowest-value activity that PMs do.

Don't mistake coordination for collaboration. Collaboration is creative problem-solving where diverse perspectives lead to better outcomes. Coordination is logistical—it's the overhead of ensuring everyone has the right information at the right time.

PMs spend an enormous amount of time sharing context across specialists (designers, researchers, and engineers) who work in silos. These specialists understand their domains well but lack empathy for each others' work. The engineer knows his domain front-to-back but doesn’t really get what the researcher does.

So, the PM works diligently through documents, decks, status updates, and stand-ups to plug the leaky holes in the system. But along this journey, context is invariably lost and important details are dropped, misunderstood, or ignored.

This coordination is a tax on Product Managers. The true "magic" that separates adequate PMs from exceptional ones is their ability to generate and validate on the right opportunities for the business to pursue, then build and execute them them. Coordination is a means to this end, not the end itself.

Nowhere is this coordination tax felt more acutely than in product discovery.

The Joy of Discovery

One of the most fulfilling aspects of Product Management is product discovery: figuring out what to build next.

Product discovery often starts from a kernel of an idea and blooms into strategy, design prototyping, and user research. The goal is to determine if an opportunity is worth pursuing and, if so, how to execute it.

The business wants PMs to do Product Discovery so that we can find new, promising ways to generate value.

But product discovery is challenging because it requires an extra-large payment of the coordination tax. It's the perfect storm for team dysfunction:

• It's always happening but always different, making standardization difficult
• We want to move fast but must coordinate across specialists working in silos with differing information and agendas
• We begin with more questions than answers and want to explore, yet product development thrives with upfront definitions of specs, timelines and resources

Many books and frameworks have been created to address these issues, but most organizations still struggle with product discovery, namely because of this coordination tax.

To remove the coordination tax, you need to remove people from the equation. But how can you do that when good product discovery needs a diverse team across product, research, design, and engineering?

AI: The Answer That Actually Makes Sense

AI is supposedly the answer for everything, but in this case, it actually is. Artificial intelligence tools are becoming increasingly effective across specialized domains, including product strategy, research, design, and coding.

Instead of spending weeks coordinating across teams, a PM can use AI tools to design, code, and run user research on high-fidelity proof-of-concepts. These PoCs don't need to be "perfect"—they just need to be good enough to validate with user feedback.

And what's more empowering? Anyone on the team can fill this "product generalist" role, whether they're currently a designer, researcher, engineer, or beyond. Their job will be to use their good taste and product sense to generate the best opportunities, orchestrate inputs and outputs between tools, and make the right decisions along the way.

Why Product Discovery is Perfect for AI

Product discovery is ideally suited for AI tools for several reasons:

1. Medium-fidelity is good enough for directional insights. AI tools can't create production-ready applications (yet), but medium-fidelity experiences provide sufficient directional insights at the discovery stage.
2. AI helps us maximize "at-bats". Traditional product discovery is constrained by time and resources, forcing teams to be precious about what they pursue — usually the most obvious idea or the CEO's preference. AI lets us explore multiple ideas, increasing our chances of finding breakthroughs. It can give PMs enough at-bats to have a chance at hitting a home run.
3. Generation is AI's sweet spot. AI can rapidly generate countless ideas, most of which won’t be good. But that’s where the PM’s expertise comes in. AI can help skilled PMs think beyond current assumptions in a systematic way.

The AI-Powered Product Discovery Toolkit

If you're a Product Leader feeling overwhelmed by all this, don't worry. Before we dive into change management, let’s start with an overview of how this new system might function.

The good news is that you can start AI-powered product discovery with just three types of tools: a product strategy tool, a prototyping tool, and a user research tool.

1. Product Strategy Tool

A product strategy tool helps PMs develop coherent strategies that can be easily interpreted by prototyping tools. It should:

• Define high-potential opportunities by synthesizing data from customer feedback, company strategy, and market research
• Help flesh out these opportunities by defining problems to solve and value to deliver
• Strengthen the strategy by helping the PM identify assumptions and blind spots

General-purpose LLMs like Claude, ChatGPT, or Perplexity work well here, or you can try specialized tools like ChatPRD.

2. Prototyping Tool

Agentic prototyping tools build working web applications from natural language prompts. They take the place of Figma for rapid prototyping. A good prototyping tool will:

• Enable non-technical people to create working applications by describing what they want the AI to build
• Allow you to publish these applications for people to interact with, just as they would interact with a website, so you can get quick directional feedback

Tools like Lovable, v0.dev, and Replit are excellent solutions to try here.

3. Research Tool

A research tool collects high-quality feedback on your prototypes and synthesizes data into actionable insights. It should:

• Present prototypes to users and run interviews to collect feedback
• Capture multi-modal user data, including text, voice, and screen recordings
• Synthesize insights into a clear format for iteration

The key is speed and volume— instead of taking weeks to interview 5-7 customers about one prototype, you could interview hundreds in just a few days.

We're biased, but we believe our tool, HEARD, is the best solution here. HEARD uses AI to conduct user interviews on everything from Figma prototypes to live web applications created by tools like Lovable or v0.dev.

Up next: From Product Strategy to Working Prototype

In my next post, I'll dive deeper into creating a product strategy and using it to build a web application with an agentic coding tool.

‍