Howie Liu

A tweet with wrong Airtable numbers went viral, was echoed on All-In, then corrected, showing how sensational misinformation spreads quickly.

Airtable doubles-down on PLG by making its homepage ask users to 'just start building', letting them experience the product immediately.

Airtable created an ‘LLM map-reduce’ that chunks large data sets, runs separate LLM calls, then aggregates results to bypass context limits.

He set a standing weekly sprint check-in on AI execution to keep half the org operating with urgency like an AI-native startup.

Howie tells PMs and designers to proactively pick up engineering and AI skills, stressing brains are malleable and resources abundant.

Howie urges teams to chase the many “watermelons on the ground” first—high-value, low-effort AI opportunities—before climbing harder trees.

Creating small, fun weekend projects forces deeper hands-on use of new AI models and reveals product form-factor possibilities beyond ChatGPT’s default interface.

Each PM, engineer, and designer should be minimally competent in the other two disciplines, then deepen their own specialty.

Airtable lets AI agents compose business apps from robust no-code primitives, acting like a domain-specific language that avoids raw code and cuts bugs.

He highlights Google, Apple, and Stripe as examples where technical, design, and product skills overlap within roles.

Howie tells his Field Agents pod to explore multiple feature directions and build quick prototypes instead of waiting for prescriptive instructions.

Howie argues the best way to deliver AI value is letting anyone self-serve and experience it, citing ChatGPT’s open sign-up as proof of superior PLG reach.

Airtable moved from feature-owned teams to business-unit pillars so groups pursue holistic outcome goals rather than incremental surface tweaks.

Make every function more full-stack—AEs demo fluently, marketers run ads end-to-end—reducing dependencies and increasing outcome ownership.

Echoing Brian Chesky, Howie argues a product-first company needs its CEO to act as chief product officer, continually driving major product reinvention.

Airtable shifted so the AI agent is now the default way of doing everything, turning the old UI into an artefact the agent manipulates to drive builder-led adoption.

Howie says fast thinking work needs people who operate with autonomy, are entrepreneurial, and can think full-stack about ambiguous product and user experience problems.

He contrasts passion-driven entrepreneurs with purely financial ones, arguing great product-centric companies are led by builders who love the product.

Howie believes CEOs operating as individual contributors, deep in meaningful details, spark non-incremental outcomes without slipping into micromanagement.

Howie views product UX as defining what the product should do and how it behaves, arguing AI tools undersell themselves with poor metaphors and affordances.