Why Airtable’s founder is betting on teams of AI agents
Howie Liu argues that SaaS economics are rooted in a fading era, so he sold Airtable and is betting Hyperagent on teams of AI agents that small businesses can employ like extra hires.
Watch on YouTube ↗Charlie's summary
- ▶ 0:29
Companies are built at the start, middle or end of a tech era, and Airtable was no longer at the start
Howie thinks Airtable was founded at the beginning of the PLG and browser-app era. AI rewrites pricing power, distribution and economics. SaaS rests on seat-based pricing and employee productivity, whereas AI means variable token spend and automated processes. Airtable was near half a billion in revenue and still growing, but he saw a fork: run it as the best SaaS business, or sell it and bet on an AI-native future.
- ▶ 6:08
Capital alone doesn't win a zero-to-one product, so Hyperagent was spun out rather than funded by Airtable's cash
Howie says capital isn't the main constraint in finding product-market fit. Overfunded startups and deep-pocketed incumbents like Google with its Plus product failed to brute-force their way in. Money can become a crutch that excuses skipping the hard detail work. AI businesses also have different margins, and Hyperagent targets SMB owner-operators rather than enterprise, which allows faster iteration. He doesn't think incumbent cash or distribution helps much here.
- ▶ 12:47
Selling beat holding on: not selling was also a risky bet, and the preference stack decides what common holders get
Howie says Airtable sold below its peak valuation, but he judged that waiting could bring a worse price as AI pressures SaaS multiples. Because most of the $1.4 billion raised was still in the bank, Airtable cleared its preference stack and returned money to common shareholders, which he thinks many SaaS companies couldn't do. His heuristic is to preserve optionality: a headline valuation isn't apples to apples, and spending a mega-round means you must earn it back before common sees a dollar.
- ▶ 18:26
Hyperagent began as a holiday of building with coding agents, which led to “agents to build agents”
Howie fell back in love with building over last winter's holidays using Claude Code, Cursor agents and Opus 4.5. For the first time agents made things that actually worked structurally, like a good engineer's output, as a solo builder. He adopted Benchmark's inside-out thinking, which starts from a new technology's capability rather than a market. Since Airtable is software to build software, he asked what if a product were a way to build frontier agents.
Give your email to unlock the rest of the key points — for every episode.
Do more with Charlie
Free — sign in with your email, no password.