From Job Description to Product Vision in 48 Hours
Recently, I was applying for a product experience leadership role at Wyze. The job description called for an interesting mix of product, design, and engineering requirements… and that left an interesting question unanswered: what exactly did the company need? A product leader? A design-savvy product thinker? Someone equally comfortable with experience strategy, emerging technology, and hands-on execution? Rather than trying to explain which version of that person I might be, I decided to treat the hiring process as a product challenge and show how I bring all three together.
For me, this conflation of skills fits right into how I’ve been working with AI. My Product experience brings the customer perspective, the rigor around articulating and prioritizing requirements, my Designer heart brings the empathy and experience that will engage users, and my Builder experience brings the real world constraints to front-end development. AI helps me do all of this now.
Understanding the problem… and the human need
I was given a recent Wall Street Journal article by the recruiter about (often humorous) false-positive notifications that home monitoring solutions had sent to their owners. I was asked to answer some questions about it. As I read the article, I was struck by the fact that the emotional impact of these erroneous notifications was not even mentioned. That was my moment of insight that carried over into how I was going to approach everything: all alerts — whether accurate or not, whether urgent or not — were being sent the same way. And if you’re receiving an alert from your home security system, your first response to seeing that alert is going to be, “Oh shhhh-oot” (along with a spike in your cortisol) even if it’s an alert for something important, but not urgent. A lack of confidence in what was being communicated with nothing to assuage the emotional response is not a good basis for building trust. Notification fatigue is a real thing. People don’t need more alerts, they need more understanding.
Everything is storytelling
I treated the interview exercise like a real product problem. I started with the job description and the Wall Street Journalarticle, then went deeper: Wyze’s existing products and web experience, its NBD and Descriptive Alert features, customer commentary, competitive products, and research into notification fatigue, interrupted attention, alarm response, and what happens when systems cry wolf too often.
A pattern emerged. The problem wasn’t simply that AI sometimes gets detection wrong. It was that an uncertain observation could suddenly become a very certain-sounding, emotionally consequential interruption. A reflection becomes a tornado. Brake lights become a house fire. The technology isn’t just making a classification mistake; it is creating anxiety, asking the customer to investigate, and slowly training them not to trust it.
That led me to a different way of thinking about the opportunity. Wyze already knows how to recognize people, pets, packages, vehicles, and motion. Recognition is not the same as understanding. What if those individual observations could be connected across cameras, devices, rooms, and time into a coherent story?
Instead of another alert saying Dog detected, Wyze could understand that Buiter left the upstairs couch, walked downstairs, passed through the kitchen, and eventually curled up on another couch. Ask, “Where’s my dog?” and the system could answer instead of making me search through clips.
Or, more importantly, it could notice when the story doesn’t quite make sense: the dog went into the bathroom, the door closed, and he never came back out.
The opportunity wasn’t simply better computer vision. It was giving the home enough context to understand what was happening, reduce uncertainty, and communicate only what actually mattered.
Designing for Trust
The harder design problem was not simply making the AI smarter. It was deciding how confidently the system should communicate what it thinks it knows.
I separated four signals that traditional smart-home alerts often collapse together: severity, urgency, confidence, and relevance. A package delivery may be highly certain but not urgent. A possible fire may have limited evidence and still require immediate attention. Treating those situations the same is how AI creates unnecessary alarm and, eventually, mistrust.
The experience therefore communicates uncertainty rather than disguising it. Conclusions can move from possible to likely to confirmed as additional evidence arrives. When confidence is low but potential harm is high, the system can notify provisionally while continuing to investigate. When time allows, it gathers corroborating evidence across cameras, audio, routines, and other permitted signals before interrupting the customer.
Just as important, the AI shows its work. Customers can review the evidence behind a conclusion, understand why Wyze surfaced it, and correct the system when it is wrong. Those corrections improve future decisions. My guiding principle was simple: do not turn probability into certainty. Trust comes not from pretending the AI is infallible, but from making its reasoning, uncertainty, and limitations visible.
Going above and beyond: Making it (feel) real.
I did not stop at strategy. I was going to build a prototype. I wrote a detailed PRD defining the experience, states, content, motion, and two core scenarios, and distilled everything into a carefully crafted “super prompt” for Lovable. To ensure the prototype would be on-brand, I captured screenshots of Wyze’s existing products and web portal, studied its visual language, and translated those references into a design.md file that gave the AI clear direction on layout, color, typography, iconography, tone, and interaction patterns.
The build produced a robust interactive prototype integrating Generative AI for the voice-driven morning digest and a conversational search experience that assembled evidence across multiple cameras to locate a pet (I call it “visual reasoning”), while retaining continuity with Wyze’s existing brand, product ecosystem, and showing how it could become more elegant, useful, proactive, and human.
This is “how I AI”
The entire process took roughly a day or two. AI accelerated research synthesis, scenario development, writing, prototyping, asset creation, and iteration, but it did not replace product judgment. The essential work, however, was human: me determining which signals mattered, identifying the most valuable opportunity, connecting customer utility to business strategy, and building something tangible enough for people to experience rather than merely discuss.
This is what AI-native product work looks like to me. It is not asking a model to generate a few screens or summarize a document. It is using AI throughout the arc of the work, from ambiguity to insight, from insight to strategy, and from strategy to a credible product experience. Building has become an extension of designing.
Executive mind, designer heart, builder hands.
This is what I do. I frame the business problem, shape the strategy, design the experience, and build enough of the future to make it real. But not just as an individual contributor. My years of people management expertise also make me a force multiplier. I can help companies and teams struggling to build in this new era.
If you need help, reach out. I’d love to discuss how I can help you make this kind of progress, too.
Turning complexity into clarity, alignment, and action.