How AI Is Exposing the Flaws in Product Development

The latest research from Harvard Business Review Analytic Services (sponsored by Miro) has a clear message for product leaders: our workflows, not AI, are the real bottleneck. In this post, I unpack the report’s findings (I’m quoted five times!) and share how to turn these uncomfortable truths into positive change.

The Alarming Stats: Everyone Knows It’s Critical – Few Are Succeeding

The survey of 332 product leaders finds that 87% consider effective product development mission-critical. Yet only 34% say their current processes are highly effective. In other words, nearly all agree product is vital, but almost two-thirds think their own approach is broken.

Meanwhile, 83% of companies expect AI to significantly improve product development. They’re planting a lot of hope in AI. But if the process is flawed to begin with, more code and chatbots won’t bridge that 53-point gap between belief and reality.

Key takeaway: Acknowledging these stats should be humbling. If our organizations admit product dev is critical yet ineffective, it’s a systemic issue — not something a tool can fix by itself.

AI Hype vs. Reality: Automating the Wrong Things

The research shows how companies are currently using AI. The top immediate benefits sought include faster time to market (44%) and quicker iteration loops (40%), along with cost savings and quality improvements. But look at what they’re automating: 39% use AI for creating charts and marketing materials, 35% for generating code, 33% for brainstorming ideas. In other words, AI is mostly being applied to discrete tasks.

As one of my quotes explains: “Data is a huge problem… without data, everything you do is just gut feeling.” Similarly, I pointed out that companies do want to run experiments, but many stop because “experimentation can take weeks to get to a result.” Applying AI to expedite tasks within a broken process won’t fix the lack of data or slow decision-making underneath.

Key takeaway: Think beyond tasks. AI can automate ticket creation or design tasks, but if the pipeline is misaligned, those boosts just amplify problems. We need to automate smarter work (like continuous user testing), not just faster work (like PRDs).

The Human Factor Remains Crucial

Even in an AI-driven future, human insight is irreplaceable. The report cites that AI can help analyze data, but “even the most advanced system can’t replace person-to-person conversations”. I reinforce that point: spending a day with users and understanding their problems “for a long time, will remain part of product [development’s] heart.” This is a powerful contrast: analytics engines vs. empathy interviews.

Key takeaway: Use AI to augment human creativity, not replace it. Maintain empathy and customer contact as the core of discovery.

Rethinking the Cycle: Steps Forward

Rather than layering AI on top, we should use this moment to fundamentally redesign product processes. In the report I urged: “Look at, from the ground up, how your product development cycle works. Where are the flaws? Where are the problems?” AI should be the reason we ask these questions and overhaul outdated practices.

Here are my recommended shifts:

• Move to Continuous Discovery: Replace rigid roadmaps with ongoing customer discovery (experiments, interviews) and adapt on the fly. Use AI to gather insights faster, but act on human feedback.

• Break Down Silos: Treat product development as end-to-end from vision to customers. The report shows poor cross-team collaboration is a top challenge. Use collaborative tools (50% of companies use visual collaboration platforms, 81% of them find it critical) and AI to foster communication.

• Outcome-Focused Metrics: Define success by user/business outcomes (engagement, revenue) instead of output counts. The report notes 60% waste or inefficiency and 52% failing revenue targets. With AI help, close the loop between features and outcomes more quickly.

Key takeaway: Don’t view AI as a silver bullet. See it as a lever to rebuild: continuous discovery, true cross-functional teams, and measurable outcomes. That’s how you make AI transformational.

Key Takeaways

• Face the data: 87% vs 34% shows how many know we’re falling short. That gap is your hook.

• Use AI to rethink, not just respeed: Quote from report: “The product development cycle should be adjusted to AI, not just have it layered on top.”

• Prioritize humans + data: Maintain customer closeness and clean data. “Gut feeling is always worse than looking at data.”

• Test and measure: Track outcomes over outputs. If an experiment fails, you’ve learned something. AI is only as good as your questions.

• Act now: The survey is a wake-up call. Use AI as a forcing function to innovate your process.

(Download the report to read more about these insights.)