In this episode of Product Unplugged, host Mike Belsito sits down with Aman Khan, Head of Product at Arize, to unpack what it really takes to build AI-native products from the ground up. From zero-to-one experiments to reaching product-market fit, Aman shares the pivotal moments, bets made, and the messy middle that defines modern product development in the generative AI space.
Key takeaways
- Product managers must evolve: “I think PRDs are probably dead in the next couple years… PMs will show up with clickable prototypes instead.”
- Don’t build in a vacuum: Customer interviews (even via LinkedIn cold DMs) were critical in shaping Arize’s early product direction.
- Product-market fit isn’t loud: “You know you have something when people just start using it… and telling their friends—without you trying.”
- Dashboards won the day: Arize’s most-used feature wasn’t what they expected—it was basic dashboards.
- Betting on GenAI was non-consensus: Their early call to shift focus to LLM evaluation was considered bold—but it paid off.
- Speed + learning = edge: Early MVPs helped them ship fast, gather feedback, and rapidly iterate—even if the first version didn’t land.
- Founders should ask: “Once you ship something… do you actually know if it’s working the way you expect it to?”
Chapters
- 00:00 – Intro & Meet Aman Khan
- 01:45 – Aman’s background: from self-driving cars to GenAI
- 04:40 – The pivot moment: ChatGPT enters the chat
- 08:30 – Early bet: Will PhDs use chatbots to build models?
- 10:00 – Starting from scratch: Zero-to-one product development
- 13:30 – MVP mistakes & learnings from v1
- 16:00 – Messaging matters: Selling the value, not just the tech
- 18:30 – What product-market fit actually feels like
- 22:00 – Unexpected insight: dashboards became the killer feature
- 24:00 – Advice for aspiring GenAI builders
- 27:00 – The last word: do you really understand your product’s impact?
