AI prototyping gives you superpowers as a PM.

But most people are putting out slop:

I sat down with former Head of Product on LinkedIn Sales Navigator Sachin Rekhi to break down how not to.

Check it out:

🎬 Youtube:youtu.be/74kamD2jUng

🎧 Spotify:open.spotify.com/show/7…

🍎 Apple:podcasts.apple.com/in/p…

It's a complete masterclass in AI prototyping for 2026.

✍️ Here were my favorite takeaways:

1. Why AI Prototyping matters

Before AI prototyping, the average product team would stay in the product space throughout planning and early PRD development. Getting into the solution space with design resources was too expensive until later. Now with AI prototyping bringing the cost close to zero, you can get into the solutioning space much sooner. This approach works - there's a reason Apple always does it. You get into the details.

2. But AI slop is real

The problem with AI prototyping is that it's almost too easy. You can easily whip up something that looks okay on the surface in 60 seconds. Unfortunately, this is a trap. The prototypes that give you the most value are going to be consistent with your product's design system and functional to the product you're building with live data.

3. There's a 15-skill mastery ladder

You want to build up all of the following skills to become great at AI prototyping:

Tools

Editing

Diverging

Prompting

Versioning

Limitations

Debugging

Product shaping

Technical editing

Executive reviews

Design consistency

Customer validation

Engineering handoffs

Functional prototyping

Designer collaboration

(all broken down in the full episode)

4. Design consistency is critical

It's so easy to match to your product's design systems these days. There's no excuse not to. In most tools, you can simply import a design system. Or better yet, work with a designer to build a base template that every future PM prototype can reference. Then the entire design system is available quickly. Bonus points for defining in something like Tailwind.

5. Diverging is the superpower it gives you

Prototypes used to be expensive. Now they're nearly free. The superpower here is you can easily diverge to test out very different solutions to the same problem. Tools like Magic Patterns have a feature built into them to do this. If your specific tool doesn't, even a prompt like "Explore multiple designs" can get you pretty far.

6. Functional prototypes increase the insights

If you go to the effort to actually connect the LLM API, use real data, add analytics, build surveys into the product, get heatmaps, and watch session recordings, you're going to get much more value from the tool than not.

🏆 Thanks to our sponsor - Reforge Build: AI prototyping built for product teams -reforge.com/aakash

What's your favorite AI prototyping tool?