PMs Who Use AI Like This Will Outrun the Rest

Last month, I ran a hands-on workshop for Mind the Product, showing how product managers can use a few AI tools to build complex, working prototypes, without writing a single line of code, and do it all in under 10 minutes.

Before the session, many PMs shared a common fear: "Will AI replace us?"

But by the end of the workshop, their mindset had completely shifted.

They realised the real competition isn’t between humans and AI — it’s between PMs who know how to use AI effectively and those who don’t. And once they saw what was possible, they felt empowered, not threatened.

That’s exactly what I want to share in this article: a step-by-step guide to help you equip yourself with the AI knowledge and tools that will fast-track your product work, and make you one of the PMs who outruns the rest.

Tooling First: Choose the Right AI for the Job

Depending on what kind of output you're aiming for, different tools shine in different areas.

Here’s a quick flow to help you choose the right AI for the task:

Once you’ve chosen your tool, the real value comes from practice. So let’s get hands-on.

"The real competition isn’t between humans and AI — it’s between PMs who know how to use AI effectively and those who don’t."

Yuan Deng, Senior Product Manager @ MHR

Building a 0-1 product prototype in 10 Minutes

To demonstrate just how fast you can go from idea to working prototype, let’s build a mini web app using Lovable.

You’ve probably heard of borrowmydoggy.com, where dog lovers can borrow dogs for companionship while their owners are away.

Now imagine the same concept... but for cats.

We’ll call it: BorrowMyCat.

Here’s the prompt I used in Lovable:

I want to build a similar website like https://www.borrowmydoggy.com/ but for cats. That allows cat owners advertise their cats on the platform for borrowing when they go on holidays, and people interested can register and apply to borrow a cat.

And here’s what Lovable generated:

Pretty impressive, right? That’s the power of giving AI something tangible, a real website it can learn from. Lovable recognises patterns from existing sites, which allows it to generate usable, well-structured UIs in no time.

So what if there’s no actual website/app that AI can go learn?

You’ll need to structure your thinking before prompting.

This is where ChatGPT (or Deepseek) becomes your planning buddy to figure out things like

• What’s the MVP?
• How many pages/tabs are needed?
• What’s the user journey?
• How do the pages link together?
• What core functions do I need?
  

When building prototypes with AI, think of it like creating a building.

You don’t start by choosing the wallpaper. You plan how many floors you need, what rooms go where, and how they connect — before decorating.

Once you’ve sketched the architecture of your app with GPT, use that as the base prompt for your prototyping tool. Make sure you also provide enough context about the project — such as the target personas and the specific pain point your app is designed to solve.

Adding Functionality: The Login & Auth Journey

If you try clicking around the Lovable prototype, you’ll notice something: although the tabs are interactive, the login and account creation journeys don’t actually work yet.

That’s because we haven’t connected the project to a database.

To make it functional, we need to integrate with a backend — and for that, I recommend Supabase.

Supabase is a backend-as-a-service platform (think Firebase, but open source and Postgres-based). It lets you manage your database, authentication, and APIs easily.

Here’s what you do:

• Create a Supabase project & connect to Loveable

• Tell Lovable (in plain English!) that you’ve connected to Supabase and you want to enable user sign-up, login, and account creation. Here’s the prompt that I used:

 ‘I am connected to the superbase. I want you to create database for me so users can register and registered user can login. Once someone register, we need to trigger email for them to confirm and they needs to be redirected to the login page of borrowmycat after they have confirmed their email’

Lovable then updates the code — and just like that… It's working!

You’ve now built a functional app that allows new users to register and log in — in under 10 minutes.

Not only that, once you hit publish, you can instantly share your working prototype with others

What This Means for You as a PM

You no longer need to wait for a designer to wire up Figma flows, or for engineers to build a prototype so you can get feedback from customers & stakeholders. You can now build working, high-fidelity prototypes in minutes using natural language.

With tools like Lovable, V0.dev, or Bolt, you can even import your Figma files and bring them to life, turning static designs into interactive demos that feel like real products.

So, if AI is this good, wouldn’t it just replace product managers? The answer is no. During the workshop, we ran an exercise where I asked everyone to design an "enhanced filter" feature for Airbnb. The PMs who simply asked AI to build it without much thought got very different and less effective results than those who approached it with customer thinking first.

For example, in the AI-led group, the filters were placed in locations that didn’t match natural user flows. Meanwhile, the PMs who first mapped out the customer journey gave the AI much clearer instructions, resulting in a more intuitive and user-centric design.

This exercise made one thing clear: the quality of AI output depends entirely on the quality of the prompt and crafting a great prompt requires product thinking. That’s exactly why AI isn’t replacing PMs, instead it’s becoming a powerful tool for PMs who know how to use it well. Think of it less like a replacement and more like a force multiplier.

The real game-changer isn’t whether AI exists, but who knows how to use it well. PMs who can harness AI to speed up discovery, prototyping, and decision-making will move faster, communicate clearer, and ship smarter. That’s the new competitive edge.