How I AI: Marily Nika’s AI Native PM Workflow (Perplexity, Veo, v0, Notebook LM)

In this episode, I was so excited to sit down with Marily Nika, an AI Product Lead at Google and the founder of the AI Product Academy. Marily is an expert in what she calls being an “AI-enhanced PM,” and she has honed an incredible workflow that radically accelerates the product development lifecycle.

Marily walks us through a tangible example any PM can copy: building a better smart fridge. The idea was sparked by a viral LinkedIn post she made after her own fridge told her that her Coca-Cola was 80 days old and expiring soon. That single, surprising user experience became the catalyst for a full-blown product concept, and she uses it to demonstrate a multi-tool workflow that takes an idea from initial research to a stakeholder-ready video prototype in about the time it takes to drink a cup of, well, Coca-Cola.

First, she’ll show us how she does deep user research in minutes using Perplexity to mine Reddit discussions and even stage an AI-powered debate. Then, we’ll see how she uses a custom GPT to instantly generate a comprehensive Product Requirements Document (PRD). From there, it's on to building a clickable prototype with v0.dev and, finally, creating a persuasive vision video with tools like Flow and Sora. We also dive into a second, fascinating workflow where Marily uses NotebookLM as an AI judge for her bootcamp's demo day.

This episode is a masterclass for any product manager, builder, or leader looking to integrate AI into their daily work. Marily's approach of “tool hopping” shows that by combining the strengths of specialized AI tools, you can achieve a level of speed and quality that was simply impossible before.

Let's dive in!

Workflow 1: The 20-Minute AI-Enhanced PM Cycle

Marily's primary workflow chains together multiple AI tools to compress weeks of traditional product work into a single 20-minute session. It all starts with a real-world insight and ends with a compelling vision that can be shared with any stakeholder.

Step 1: Instant User Research with Perplexity

Before writing a single line of a PRD, a good PM validates the idea and understands the user's needs and concerns. Traditionally, this means weeks of user interviews or expensive agency work. Marily does it in about three minutes.

Tool of Choice: Perplexity, a conversational search engine.

The Magic Feature: Marily uses the 'Focus' dropdown and selects Discussions. This narrows the search to forums like Reddit, giving her direct access to raw, unfiltered user opinions.

Initial Prompt: She starts with a simple query to gauge general sentiment.

would families be interested in a smart fridge?

The Debate Technique: This is where it gets brilliant. To avoid the typical AI bias of just agreeing with your idea, Marily prompts Perplexity to create two opposing agents and have them debate. This surfaces the strongest arguments for and against, revealing the critical features needed to win over skeptics.

Create two agents, one that is pro smart fridge and one that is against smart fridge. Use everything you read and have these two agents debate at least, I dunno, like 20 times about it, and give me the minimum set of features I would need in order to convince the against agent.

Outcome: Instead of a generic summary, Perplexity provides a 20-round debate synthesized from real user discussions, and most importantly, it outputs a minimum set of features required to achieve product-market fit. This includes things like local-first intelligence for privacy, which becomes a core tenet of the product.

Step 2: From Research to PRD with a Custom GPT

With a validated feature set in hand, the next step is to create a Product Requirements Document. Marily has streamlined this by building her own custom GPT in ChatGPT.

Tool of Choice: A custom GPT pre-configured with her preferred PRD template and writing style.

The Process: She performs what she calls “tool hopping.” She copies the minimum feature set generated by Perplexity.

The Prompt: She then pastes this directly into her custom GPT with a simple instruction.

generate a PRD about a smart fridge that has these features

Outcome: In about 90 seconds, her custom GPT generates a well-structured PRD. It includes a problem statement, target users (privacy-conscious consumers, DIY enthusiasts), feature analysis, and prioritization. It's not perfect, but it provides an incredible head start, allowing her to focus on strategy rather than formatting.

Step 3: Bringing the Vision to Life with v0 Prototyping

A PRD is great, but a tangible prototype is far more powerful for communicating a vision. To bridge this gap, Marily hops to her next tool.

Tool of Choice: v0.dev, an AI-powered tool that generates UI from text prompts.

The Process: She copies the entire generated PRD from ChatGPT.

The Prompt: She pastes the PRD into v0 with a straightforward command.

Create the UI of a smart fridge, given this PRD as an input

Outcome: v0 generates a surprisingly detailed and interactive dashboard for the smart fridge. It includes widgets for temperature, power usage, and my personal favorite, a Door Status widget—a perfect feature for parents! Because the PRD emphasized local processing and privacy, the prototype even includes a 100% Local Processing badge and system diagnostics showing an onboard GPU. This prototype is now a powerful asset for product reviews, allowing stakeholders to see, touch, and feel the product vision.

Step 4: Selling the Dream with AI-Generated Video

For the final step in stakeholder influence, Marily creates a short promotional video to showcase the user experience.

Tools of Choice: Flow (by Google Labs) and Sora.

The Prompt (for Flow): She again uses the core feature list as the input.

create a promotional clip for two, for a couple, using a smart fridge that has these features

The Results & Iteration: The initial results from Flow were interesting but slightly off—one version had a screen inside the fridge, and in another, the man in the video mysteriously transforms into another woman. This highlights a key skill for AI-enhanced PMs: iteration and prompt refinement.

The Improved Version (with Sora): To get a more compelling result, Marily switched to Sora and used its Cameo feature to insert herself and Mark Cuban into the scene. This version was a huge leap forward, generating a short clip complete with dialogue that actually explained the product's value. The AI even had her mention a “Greek salad,” which was a fun, personalized touch since Marily is Greek.

Outcome: This final video clip is the perfect cherry on top. Paired with the PRD and the interactive prototype, it creates a comprehensive package that powerfully communicates the product's vision, functionality, and user value in a way that a slide deck never could.

Using NotebookLM as an AI Judge for Product Demos

Beyond her day job, Marily runs the AI Product Management Bootcamp, which culminates in a demo day where students pitch their products to VCs and judges. To add a unique and impartial voice to the judging panel, she turned to another powerful AI tool.

Tool of Choice: NotebookLM, a research and writing assistant from Google.

The Setup: Marily records the audio from every student's pitch and uploads the individual audio files as sources in NotebookLM.

The Instructions: She then gives NotebookLM a clear set of instructions to act as a judge. She tells it to select the top three apps based on specific criteria:

• Innovation
• Impact
• Storytelling

The Magic Feature: NotebookLM doesn't just output text. It can generate an 'Audio Overview,' creating a podcast-style summary of all the pitches. Even cooler is the interactive mode, which simulates a live radio call-in show with two AI hosts. During her bootcamp's demo day, Marily can literally interrupt the AI hosts and ask them to announce the winners on the spot.

Outcome: This creates an incredibly engaging and fun experience for the students. The AI acts as an unbiased judge, and its analysis is surprisingly accurate, often aligning with the human judges' choices. I thought this was a brilliant application, and it immediately made me think of other use cases, like judging sales pitch competitions or hackathon demos.

The New Era of Product Management

What we witnessed in this session was a complete reimagining of the product management process. In the time it would traditionally take to schedule the first user research interview, Marily completed the entire cycle from idea validation to a compelling, multi-asset pitch package.

Marily’s philosophy of “tool hopping”--using the best specialized tool for each step of the job--is a powerful takeaway. By mastering a handful of these applications, you can dramatically amplify your impact, spend more time on high-level strategy, and communicate your vision with more clarity and persuasion than ever before.

As Marily so aptly put it, "PMs that use AI are the ones that are gonna take over the role of people who don't use AI."

So, I encourage you to try this workflow. Pick an idea, fire up Perplexity, and see how far you can get in 20 minutes. The future of product management is here!

Thanks to Our Sponsors

• WorkOS — Make your app Enterprise Ready today
• Miro—A collaborative visual platform where your best work comes to life

Episode Links

• YouTube
• Spotify
• Apple Podcasts