Product Craft is the fractional product leadership practice of Roman Levin, a fractional CPO who embeds with B2C and D2C startups 2–3 days a week to own product strategy, roadmap, design direction and engineering execution. AI-native by default, building with AI agents rather than only advising on them.
A fractional CPO is a Chief Product Officer who works with your company part-time, usually two to three days a week, on a monthly retainer. The job itself is the same as the full-time one. I own the product strategy and the roadmap, work closely with your designers, run prioritization, lead the product and UX work, and sit inside the engineering and leadership conversations where the decisions actually get made.
What changes is the overhead, not the seniority. A company past 200 people usually needs a CPO in the building every day. A startup under 50, especially a founder-led one, doesn't need a full-time CPO. It does still need a CPO. That gap is the sweet spot the fractional model exists for.
What's different from the fractional CPO next door is that AI isn't a service line here, it's the default. There's no "AI practice" to bolt onto an engagement. Every roadmap I own gets built by someone who ships with agents daily and knows where they break. That's usually what keeps a team from bolting a chatbot onto the side of the product, or spending two quarters on a demo that never ships.
There are three ways to work together, from running the product function outright to monthly mentorship.
See the engagement modelsDelightful products need a mix of disciplines working together: customer empathy, human psychology, deep technological expertise, great design and data you can trust. A subset of them produces an OK product. All of them produce an exceptional one, and executing well takes more still.
Deep customer empathy is the secret weapon of great product leaders. Caring about your customers, getting into their shoes, talking with them and feeling their pain is what produces products people care about back. I love talking to customers, and the intuition it builds is the difference between a product that works and one that lands.
People don't behave the way specs assume. They avoid loss more than they chase gain, they take the default, and they abandon at the moment the effort stops feeling worth it. Most conversion problems come down to an unclear path nobody wants to walk, or one choice too many, rather than a missing feature.
Knowing what is expensive to build changes what you choose to build. Having shipped software as a developer and a CTO, I can tell a two-week problem from a two-quarter one, push back with engineering on the substance, and design AI-native products around what the technology actually does well rather than what the demo suggested.
A working product is not the same as a delightful one, and delight is what brings customers back and gets them talking. That takes a design eye, deep UX understanding, customer empathy and a sense of what works in the hand, not just a spec that got built.
Opinions are cheap and mine are no exception. I believe in data, and in making sure the data is trustworthy before anyone leans on it. We define KPIs and OKRs that mean something, agree how they get measured, and then use them to make decisions instead of arguing from taste.
Roman Levin has spent 20 years building consumer and tech products: as a senior product leader at high-growth startups and global consumer brands in the US, as a software engineer and CTO before that, and as a founder throughout. He has built product with teams at Sunbit, Shutterfly, CLEAR, easyplant, FlowRx, Fignel, Linx, WODOX and many more, across e-commerce, marketplaces, fintech, identity and AI-native products.
That range is the whole point. I've built products from the ground up and scaled them to millions of users, and I've worked with everyone from two-person founding teams to CEOs of multi-billion dollar companies. As a founder myself, I know what it takes to build something people actually love, and what it costs when you get it wrong.
I build with AI, not just talk about it. My own company runs on AI agents across marketing, operations and finance, which is where I learned where agents earn their keep and where they burn money without anyone noticing. That's the experience I bring to teams trying to turn a promising demo into a product.
Bottom line: I've been the person accountable, which is a different job from advising the person accountable.
I take one or two clients at a time, deliberately. Product leadership done properly means being around and involved in the work, and that doesn't scale past a couple of teams.
A free 30-minute intro call. You'll leave with a clear read of your product challenges, ideas on how to solve them, and a sense of whether we want to work together.