Instead, the goal is product sense: enough fluency to bring disparate signals together and make informed decisions.

“You don’t have to become an expert,” she said, “but you need to become fluent enough to understand how to get to a synthesized point of view and create direction for the product.”

Think of the PM less as the person playing every instrument and more as the person who understands how the orchestra fits together.

That becomes particularly important when the tempo accelerates.

AI changes the bottleneck

Much of the conversation about AI in product management has focused on efficiency.

AI can help draft requirements, analyze information, accelerate coding and take some of the administrative grind out of product development. Julia sees real value there.

She also sees a trap.

If AI reduces the cost and time required to build software, then development capacity stops being the only constraint. The bottleneck moves upstream.

Now you need enough good ideas to feed the machine.

“If we are able to deliver things more quickly, it’s not going to be speed as your issue anymore,” Julia said. “It’s going to be the quality of your ideas, the quality of that discovery that’s going to make a difference.”

That flips the conventional AI productivity story on its head.

A team that doubles its development speed without improving customer discovery hasn’t necessarily doubled its productivity. It may simply have doubled its capacity to produce features customers don’t need.

Mike put the problem succinctly: faster development can increase “our ability to confuse and overwhelm customers.”

That’s why AI should change where product managers invest their time.

Less energy maintaining the machinery of delivery. More energy understanding the problem.

Discovery is not a phase

This brings us to product discovery, a term that sometimes gets treated as the homework teams complete before the “real” work of building begins.

Julia rejects that separation.

“Discovery is ongoing,” she said.

Before launch, that can mean generative research, evaluative research, concept testing and direct customer feedback. Once a product is in market, discovery expands to include customer, platform and business metrics alongside qualitative insights.

The common denominator is synthesis.

Teams collect mountains of information. Interviews. Analytics. Support tickets. Experiments. Research reports. AI can make that mountain even larger.

But collecting evidence isn’t the same thing as understanding it.

“Synthesis is where the action happens,” Julia said, “and where you’re actually making decisions and meaning.”

Her practical recommendation is deceptively simple: create regular synthesis sessions where the team reviews what it is learning and decides what those signals mean together.

And don’t outsource that thinking wholesale to AI.

AI can help organize evidence and surface patterns. But Julia cautions against allowing it to substitute for human judgment.

That distinction—AI as amplifier rather than oracle—may become one of the defining product management skills of the next few years.

Get comfortable being wrong

There is another skill that becomes more valuable when teams can move faster: intellectual humility.

Product people are supposed to have opinions. They need a point of view about the customer, the market and what their product should do next.

They also need to accept an uncomfortable truth.

Their clever idea might be wrong.

Julia describes the mindset as “strong opinions, loosely held,” a phrase that emerged from her interview for the book with experimentation leader Vito Carvalho.

The important word may be loosely.

Experienced product managers, Julia observed, have accumulated enough scar tissue to understand that the objective isn’t to prove themselves right. It’s to learn whether they’re right before the organization invests too heavily in the answer.

That changes the goal of product experimentation.

Instead of obsessing over creating the perfect hypothesis, experienced teams focus on learning velocity: How quickly can we test assumptions, gather customer evidence and adjust?

This is where A/B testing, concept testing, prototypes and qualitative customer research become complementary rather than competing methods.

Test earlier. Learn sooner. Spend less money discovering you were wrong.

AI only strengthens the case for shifting validation left because cheap production can tempt teams to skip the uncomfortable work of proving that an idea deserves to exist.

Zero backlog is the wrong finish line

One particularly revealing moment in the conversation concerned the prospect of AI helping teams achieve a “zero backlog.”

Imagine finally clearing that endless queue of feature requests.

Sounds wonderful.

Except a backlog isn’t a to-do list. It’s a collection of possibilities, requests, assumptions and ideas of wildly different quality.

“We need parking lots for a reason,” Julia said.

Building every requested feature simply because AI has lowered the cost of building it risks transferring the organization’s clutter directly to the customer.

More buttons. More settings. More complexity. More things to maintain.

Sometimes good product strategy means subtraction.

“If you’re not killing things, you’re probably not doing your job well,” Julia said. “We know not every feature is going to have that value for customers.”

That’s an important corrective to the AI-fueled obsession with output.

The measure of a strong product organization isn’t how much it can ship.

It’s how deliberately it chooses what not to.

The product manager as glue

Julia also challenges another familiar product management metaphor: the PM as the “CEO of the product.”

Real CEOs control budgets, reporting structures and hiring decisions. Most product managers don’t.

What they have is responsibility paired with influence.

Julia prefers another metaphor.

“You’re the glue.”

That means bringing engineering, design, data and other partners into the conversation early. It means giving them the business context behind a decision instead of treating them as a production line waiting for instructions.

And perhaps most importantly, it means arriving with a problem rather than a finished answer.

The cross-functional leaders Julia interviewed for her book were remarkably consistent on this point. They wanted product managers who were curious about their disciplines, understood enough to collaborate intelligently and didn’t simply throw solutions over the fence.

“Nearly all of the product leaders or executives cited curiosity as the single most important thing they’re looking for,” Julia said.

Her translation is better: be nosy.

Ask why. Talk to customers. Understand how the technology works. Learn what the data means. Invite disagreement. Find the evidence that could prove your favorite idea wrong.

AI can accelerate an enormous amount of product work.

Curiosity still belongs to us.

And as the cost of building continues to fall, the organizations that succeed may not be the ones that produce the most. They may be the ones disciplined enough to understand their customers before deciding what deserves to be produced at all.

As Julia put it: “It’s not about delivery, it’s about that customer experience.”

Additional resources

• Listener offer: Use code UT20 when purchasing The Product Management Playbook directly from Rosenfeld Media for 20% off.
• Julia Barham on LinkedIn
• The Product Management Playbook
• Author interview with Rosenfeld Media
• Progressive Insurance, where Julia is a Senior Director of Product Management
• Mike Mace on LinkedIn
• Nathan Isaacs on LinkedIn
• Webinar: Continuous discovery: transform your product development process — Hear how teams at Sage, Travelport and Trivago integrate continuous discovery into product development to move beyond feature delivery and build around customer outcomes.
• Guide: A product executive’s guide to deploying human insight — A practical guide to embedding rapid customer validation throughout AI-accelerated product development, reducing rework and making more confident product decisions.
• Podcast: AI can help you build faster. It can’t tell you what’s worth building. — Product leader Rich Mironov joins Mike Mace to examine why faster AI-assisted development puts an even greater premium on customer discovery, product strategy and human judgment.
• Blog: AI-powered product research: Why the fastest builders need to become the fastest learners — Explores how AI-powered product research can help customer learning keep pace with faster building while turning accumulated customer feedback into organizational intelligence.

Frequently asked questions (FAQs)