Why do so many AI and robotics projects fail?

Engineers get excited about the latest robot navigation breakthroughs. Meanwhile, business leaders keep asking:

“But where’s the ROI?”


A recently recorded Lean Startup podcast perfectly captured what I’ve seen for years as a consultant in AI and robotics.

The guest, Siddarth Anand , nailed the core issue:

🔥 There’s a massive disconnect between business leaders (focused on outcomes) and tech teams (focused on data, models, and tools).


So what’s missing?

👉 The AI Product Manager—someone who speaks both languages and bridges the gap.

Oddly, when I look at open roles in companies aiming to integrate AI or robotics, I rarely see job postings for this kind of product leader.

Instead, I see positions for ML engineers, data scientists, or traditional PMs—with no mention of lean experimentation, hypothesis testing, or rapid validation.

It’s a blind spot. And it’s costing teams time, money, and trust.

For over 10 years, I’ve applied lean startup thinking to AI projects, and these principles guide every engagement:

✨ Start with customer problems, not cool algorithms

✨ Validate solutions through fast, cheap prototypes

✨ Focus on business OKRs, not F1 scores

✨ Facilitate real collaboration between product, data, and business

✨ Drive iteration from real-world feedback—not internal assumptions

Traditional PMs write specs and plan roadmaps.

Lean AI PMs test hypotheses, validate with real users, and iterate fast.

Less time in conference rooms or behind a desk, more time getting feedback that stings.

Curious—have you seen this pattern too?

What’s worked (or failed) in your experience aligning AI innovation with customer value? I am eager to connect with you and hear about your experiences. .