The work #

Starlifter was an AI-native business intelligence platform built for operators and decision-makers — the people who need answers from their data but don’t think in SQL. The premise was that the combination of LLMs and good product design could make data analysis accessible to the people who actually make decisions, not just the analysts who support them.

I was the CEO. Product direction, customer discovery, positioning, technical architecture, and — for long stretches — the primary builder. This was the full zero-to-one experience: not a side project with a landing page, but a real company with real customers, real constraints, and real consequences for every decision.

What I did #

The work was the full founder operating stack, all at once. Customer discovery — understanding what “I need to understand my data” actually means for different kinds of operators, and where existing BI tools fail them. Product direction — deciding what to build first, what to defer, what to kill. Technical architecture — choosing the right foundation when you know you’ll need to move fast and pivot. Positioning — figuring out how to explain what you’re building to people who’ve been burned by every “AI for X” pitch.

And sales, and fundraising, and prioritization, and saying no to things that seemed promising but weren’t the right bet yet.

Why it matters #

Every senior leader in tech says they can “operate in ambiguity.” Most of them mean they can manage through ambiguity — with a team, a budget, org support, and air cover from above. Founder ambiguity is different. You’re the one making the call, living with the outcome, and showing up the next morning to iterate on it. There’s no escalation path. There’s no committee to defer to. The buck stops, and then you ship again.

Starlifter sharpened everything: product instinct gets sharper when every feature decision has immediate revenue implications. Prioritization becomes ruthless when there are more ideas than hours. Technical judgment improves when you’re the one living with the architecture choices six months later.

The AI-native angle matters too. I wasn’t building a traditional SaaS product with AI bolted on — I was building from the ground up with LLMs as core infrastructure, learning firsthand what works, what doesn’t, and where the actual leverage is in AI-assisted product development. That experience informs everything I’m doing now.