Marily Nika, Ph.D

For years, I’ve been advocating not to build what’s possible, but to build what’s missing. Today, I decided to write about what’s missing: excited to announce that the AI Product Playbook is finally here! 📘🤖

My second book was born out of hundreds of hours of collaboration with someone I’ve never actually met in person, my colleague Diego — yet we’ve worked side by side, virtually, as AI Product Builders at Google. Our connection started from a simple exchange of ideas about what it really means to build ML products — and quickly turned into something much bigger.

For years, I assumed all product managers faced the same challenges I did. But AI product building is different — it’s probabilistic, data-dependent, and often ambiguous in ways traditional PM frameworks don’t capture.

That realization sparked this project.
We didn’t just want to describe what’s possible in AI — we wanted to help others discover this role, understand it and build what’s missing.

If you’re a product manager, builder, or technologist trying to navigate the new frontier of AI-driven products, this book is for you.

It is different than my O’ Reilly book in that the playbook provides the broader foundation, career path, technical deep dive and strategic thinking; but my O’Reilly book is about leaders as they (and their teams) move into execution/build mode. So I recommend both.

The AI Product Playbook: https://amzn.eu/d/6hDqlEG (EU), https://a.co/d/japctzI (US)

Building AI-Powered Products: https://amzn.to/3Dbufkp

In fact - ALL students at my upcoming November cohort course will get a free book.

Comment your interest in this post and we will send you a DM with the logistics.

Our job is shifitng, It's now time to build.

If you are a PM: Your role shifts from spec-writing to actually launching.
If you are a designer: The Figma-to-code gap is closing.
If you are an engineer: Routine coding is being commoditized, but engineers who partner with builders will become force multipliers.
For companies: Product cycles compress. The distance between idea and customer feedback shrinks.

Traditionally, I would go from:
✨ Idea → PRD → Debate → Refine → Build

Today, I go from:
✨ Idea → Brainstorm with AI → Prototype the vibe → Team experiences it → Refine → Build

That's exactly what I talked about yesterday at the Product Leader Summit, and I love that people are ready to build with me.

I am now using ONE tool and I'm assigning one engineer to every single person of my cohort so that you _actually_ leave my course with a product in production or a business ready to make money. We have instructors and mentors from Meta, Anthropic and Amazon to guide youm plus 3 certifications in one course.

🚨 🚨 🚨 Come build your side app or business with me in November: https://lnkd.in/gn3CXhjr, we have 14 spots left.

Reach out to me if you have been impacted by layoffs and need a discount or reach out to me for private trainings for your company.

Stop half-building and go from Idea → Prototype → Launch (for real) ...with me.

By now, so many of us have played with prototyping tools. You had a spark of an idea, started building and then ...stopped. Maybe you didn’t have engineering support? Maybe you got stuck? Or maybe you just weren’t sure if you were really onto something.

But that's the thing — you were. You just needed the right support system to bring it to life.

💭 If you’re not sure you’re onto something: Zoom out. Ask yourself:
- “What problem am I solving, and for whom?” Try using the problem–solution-validation loop:
1️⃣ Define the pain clearly.
2️⃣ Create one simple prototype that addresses it.
3️⃣ Get feedback from 3 people who actually experience that pain.

💡 If you’re not sure what you’re doing: Use the MVP Map:
1️⃣ Write down your idea in one sentence.
2️⃣ Identify the 1 core feature that delivers value.
3️⃣ Cut everything else. Focus on getting that one thing working end-to-end.

🧭 If you forgot your “why”:
Pause. Reconnect with the moment that gave you the idea.
If it no longer excites you, pivot or pause intentionally but not out of burnout.

💡 This November, I’m opening my next AI Product Management Bootcamp & Certification Cohort — designed exactly for you: the creators who started, but didn’t finish.

This time, every team gets paired with an engineer to help you actually launch your product. We’ll go from idea → prototype → production, together. You’ll leave with something real, live, and launched.

And we’ll celebrate with an in-person meetup in SF, where I’ll hand out free books and sign them for you.

Let’s stop half-building. Let’s ship.

If you comment "Framework", I'll send you a framework I'm putting together to help and also an offer for you or your teams/groups if you are interested in a private workshop for your company via Maven.

I loved speaking at Stanford's School of Medicine on AI tools and offering a few demos. One of the reasons I love doing this is experiencing the aha moment in people who are ready to build and create — especially when it’s about a very specific domain with pain points that only they truly understand.

Every time I offer these sessions I’m reminded that we can’t assume what people know — or don’t know. Tools that seem “everywhere” in one circle can be completely new in another. That’s actually exciting, because it means there’s still so much room for exploration, learning, and empowerment.

The best part is when someone connects an idea or a problem they care deeply about with an AI capability they’ve just discovered — the oh wait, I could use this for… moment.

Here's 7 key metrics every AI PM should track — not just to measure engagement, but to ensure your AI is useful, safe, and trusted.

Too often, we focus on DAU or churn… but, especially if you're building conversational products, you need new metrics — ones that capture meaning, depth, and trust.

Here’s the framework I use 👇

I used a pyramid because each layer supports the next: without factual, safe foundations, you can’t earn trust or scale responsibly.

-The foundation is Model Quality — your AI must be accurate, safe, and fast before anything else matters.
-Above that is Interaction Quality — can users have meaningful, multi-turn conversations that feel natural and helpful?
-Then comes Trust & Delight — do users enjoy the experience and come back because they trust it?
-Higher still is User Value — are people actually achieving their goals faster, easier, and better?
- And at the top sits Sustainability — are you doing all of this responsibly and efficiently (revenue / compute $, LTV / CAC)?

Success in conversational AI = Useful × Safe × Trusted

<><><><><><><><><><>
Follow Marily Nika, Ph.D for AI PM education, certifications and insights.