How we restructured Airtable’s entire org for AI

Howie Liu, co-founder and CEO of Airtable, discusses how he has returned to hands-on product work as an "IC CEO" and how Airtable restructured its org and product approach to compete in the AI era. The conversation covers the viral misinformation about Airtable's decline, splitting engineering into fast and slow thinking groups, rethinking the company from scratch for AI, and the cross-functional skills PMs, designers and engineers now need.

01Key learnings

• Leaders should stay close to the product details, since in AI-era products the interaction design and underlying behavior are the product itself and can't be handed off wholesale.
• Use AI products constantly, even multiple times an hour, and build small weekend projects to understand what models can do and which product form factors they suit.
• Ask whether you would build the same mission from scratch as an AI-native company; if your existing assets don't give a real edge, consider selling and starting fresh.
• Split teams into a fast-shipping group for AI capabilities shipped weekly and a slow-thinking group for deliberate, complex infrastructure work, which together fuel top-of-funnel adoption and durable enterprise growth.
• Get comfortable with evals only after you know the use cases; early on, explore with open-ended vibe testing and divergent prototyping before converging on measurement.
• Make prototypes and working demos the main artifact instead of decks or PRDs, so teams can feel how a product behaves with realistic and unrealistic inputs.
• Encourage every role to become a hybrid by getting minimally competent in the adjacent disciplines, such as PMs prototyping and designers understanding technical constraints.
• Give teams explicit permission and time to play with new AI tools, such as blocking days or weeks off meetings, to build curiosity and fluency.

02Frameworks mentioned

Summary and takeaways written by PM Atlas; quotes are short excerpts. © the original author.