How we restructured Airtable's entire org for AI | Howie Liu (co-founder and CEO)

Key Concepts:

• AI-native approach
• ICEO (CEO as Individual Contributor)
• Fast Thinking vs. Slow Thinking
• Product-Led Growth (PLG)
• Chief Taste Maker
• Maximally Accelerated
• Experimentation and Play
• Hybrid PM/Prototyper
• Growth Mindset
• Humility and Gratitude
• Eval vs. Vibes

1. Air Table's "Near Death" Tweet and Market Reset

• A viral tweet claimed Air Table was "dead," overvalued, and not profitable.
• Howie Lou states that the tweet was based on incorrect data, with revenue and growth rates significantly understated.
• The tweet gained traction after being discussed on the All-In podcast, framing it as part of a broader discussion about the valuation of decacorn companies during a market reset.
• All-In issued a correction later, revising their view on Air Table.
• Howie learned about the power of memes and misinformation on social media.

2. The Rise of the ICEO: CEO as Individual Contributor

• Howie discusses his shift back to being more hands-on with product development, coding, and UX design.
• In the early stages of Air Table, Howie was deeply involved in the technical details, which was crucial for finding product-market fit.
• As the company scaled, he moved away from these details to focus on building teams and processes.
• AI's paradigm shift necessitates a return to being in the details to continuously refine product-market fit.
• Howie emphasizes the importance of being the "chief taste maker" and actively participating in creating the product.
• He advocates for CEOs to use AI tools daily to understand their capabilities and potential applications.
• Howie proudly claims to be the highest inference cost user of Air Table AI, intentionally spending on AI to gain strategic insights.

3. Restructuring for AI: Fast Thinking vs. Slow Thinking

• Air Table reorganized its EPD (Engineering, Product, Design) organization into two groups: Fast Thinking (AI Platform) and Slow Thinking.
• The original structure had teams responsible for specific features or surface areas, leading to incremental improvements rather than holistic changes.
• A subsequent reorg divided the company into business units (Enterprise, Teams, AI, Solutions), but it still didn't allow for the rapid innovation needed in the AI era.
• The Fast Thinking group focuses on shipping new AI capabilities on a near-weekly basis, aiming for a "jaw-dropping" user experience.
• The Slow Thinking group focuses on more deliberate bets that require more premeditation, such as infrastructure changes and complex data handling.
• The two groups complement each other, with the Fast Thinking group creating excitement and the Slow Thinking group enabling durable growth.

4. The Fast Thinking Team: Archetypes and Autonomy

• The Fast Thinking team requires individuals who can operate with a lot of autonomy and are entrepreneurial in nature.
• They need to think full-stack about the problem, considering the user experience and the "wow factor."
• The team is working on a new capability that allows users to describe the app they want to build and then iterate on it with a conversational agent.
• The agent can also do code generation to extend the apps with bespoke functionality or visuals.
• This requires blending design thinking with technical constraints and understanding the capabilities of AI models.

5. The Importance of Product UX in the AI Era

• Howie considers himself a product UX person at his core, viewing UX as more than just cosmetic design.
• He believes that the product UX should represent and behave for the user, and that this is the product itself.
• He argues that many AI products under-merchandise their capabilities and lack visual metaphors or affordances to help users understand them.
• Air Table tries to show all the different states and use colors to play them up, making the product experience AI-centric.

6. Maximally Accelerated and Experiential Learning

• Howie emphasizes the need to move faster in the AI era, echoing Nick Turley's concept of "maximally accelerated."
• He believes that AI value is best delivered experientially, through product-led growth (PLG).
• Chat GPT's success is attributed to its frictionless trial experience, allowing users to experiment and discover its capabilities.
• Air Table is shifting its attention back to builder-led adoption, showing the value of AI in the product itself.

7. The Future of App Building: AI-Powered Democratization

• Howie sees a trend towards AI-powered vibe coding and app building, where users can describe what they want to build and have AI generate it.
• He believes that the form factor and product UX need to evolve with the capabilities of the underlying models.
• Air Table's vision has always been to democratize software creation, and AI is a different means to the same end.
• Air Table's existing no-code components allow it to execute better on this vision than if it had to start from scratch.
• The agent can manipulate these primitives as a domain-specific language (DSL) to build business apps, rather than writing code from scratch.

8. Learning and Experimentation: The CEO's Daily Routine

• Howie uses as many different AI products as possible, including those outside of Air Table, to understand their capabilities and form factors.
• He invents side projects to have a real reason to use these products, such as creating funny video shorts with AI-generated scripts and avatars.
• This helps him understand the models and the product form factors in which they can be placed.
• He stresses the importance of play and experimentation for his team, encouraging them to block out time to explore AI products.
• He wants to see actual interactive demos and prototypes, rather than decks or PRDs, to get a feel for the product.

9. The Evolving Roles of PM, Engineering, and Design

• Success with AI tools depends more on individual attitude and polymathism than on specific roles.
• There's a strong advantage to those who can cross over into the other two roles, becoming hybrid unicorn types.
• Designers need to be technical enough to understand how the models work and prototype concepts.
• PMs need to get into the technical details and get hands-on, rather than just writing documents.
• Overall, companies can get more done with fewer people, but this means that each person needs to be more versatile.
• PMs need to become hybrid PM/prototypers with good design sensibilities.

10. The Power of Eval vs. Vibes

• For completely novel product experiences, Howie suggests starting with "vibes" rather than "evals."
• This means testing in a more open-ended way to see if the product even works in a broad sense.
• Eval are more useful once you've converged on the basic scaffold of the form factor and know what use cases you want it to work well for.
• Eval can constrain you too early, so it's important to be divergent first and then converge.

11. Key Advice for Shifting to an AI-Native Company

• Reset expectations on pace and urgency, understanding that things move incredibly fast in AI.
• Get stuff out so that you can learn how people use it and what it's capable of.
• Encourage people to play with the latest stuff and give them time to stay on top of it.
• Rethink what you would do to achieve the same mission if you were starting today, leveraging your existing unfair advantages.
• Talk to AI constantly, multiple times an hour.
• Break down silos and encourage everyone to become more full-stack.

12. Counterintuitive Lessons and Founder Mode

• The most counterintuitive lesson is that scaling up and industrializing processes can lead to a loss of holistic thinking and magical integrative value.
• The CEO has to play a CPO role and care about the product.
• It's important to ship and learn and experiment a lot more in this era.
• Don't just trust recommendations from smart people, but understand their chain of thought and why they recommend it.
• Founder mode is about being in the weeds, being in the details, and trying things yourself, not delegating to execs.
• It's about finding the right balance of caring about the details that matter and tying them together across different groups.

13. A Decade Ago Howie's Ear

• Don't step away from the details that you love.
• Always make sure that is still your number one priority, even if other stuff has to add to your plate.
• Remember what you actually love about it and come back to that, because that's the only way to keep doing this for a long time.

14. Empowering Conclusion

• It's never been easier to learn these things, with super intelligences that you can talk to and interactive tools to build stuff.
• Everyone can learn how to be a versatile unicorn-like product engineer designer hybrid in the AI-native era.
• The only thing stopping you is going out and doing it.
• Approach each day with a spirit of humility and gratitude.