Bret Taylor is the co-founder of Sierra and chairman of the board at OpenAI. We cover his legendary experience rewriting Google Maps in one weekend, the myriad applications of AI agents beyond chat bots, and the one question every company should ask itself.
Principles & Lessons:
1) Small, empowered teams often outperform large, bureaucratic ones. Bret reminded us that “if you want to make a software project go twice as slowly, add another person to it,” tying it to the “Mythical Man-Month” concept. He recalled how a lean group built Google Maps under severe browser constraints, which forced high agility. “When you have lots of people working on a project, it requires a lot of upfront planning,” he said, whereas smaller units “don’t need to go through a committee to talk about rethinking something.”
2) Agents will become new digital interfaces for brands. Bret explained that AI agents are “essentially a system that can reason and take action autonomously,” and sees them emerging in three categories: “persona-based agents that do a specific job,” “customer-facing company agents,” and “personal agents.” He believes every brand will create its own agent the same way it once built a website: “We think in 2025, the way you will engage with your customers will be your AI agent.”
3) Effective AI solutions require factual knowledge, procedural knowledge, and system integration. In describing how Sierra helps companies launch agents, Bret said you first need reliable facts—“rather than relying on the AI…you present the model with the knowledge it’s allowed to know.” Next comes process specifics—“What’s the best approach if a Sonos speaker stops working?”—and finally “access to the underlying systems themselves.” This three-part stack helps transform “simple chatbots” into robust agents that can act on behalf of customers.
4) The shape of a company may drastically change when tasks are automated. Bret urged business leaders to re-examine how AI can alter margins and disrupt entire industries, citing what “Amazon did to many incumbent booksellers” when the Internet took off. New entrants might exploit cheaper operations or different cost structures, and Bret advised large firms to “adopt AI internally and externally,” letting employees use and master LLMs instead of resisting them.
5) Be willing to rewrite or rethink your products with new lessons. He recounted rewriting Google Maps “based on all the hard-won lessons” from its messy early code. Having the freedom to fix core architecture “in a weekend” (once he already knew where the pitfalls were) demonstrated that “when you have a small, empowered team,” you can quickly pivot to a cleaner solution rather than remain hostage to past mistakes.
6) True customer-centric thinking means direct listening at scale. Calling Salesforce “the most customer-centric company I’d ever experienced,” Bret highlighted that many firms give lip service to focusing on customers but rarely reflect it in daily ops. He shared that “deep listening” was a key Salesforce trait—something product-led companies often miss. He expects AI to make personalized conversation far cheaper, letting brands “have a direct personal conversation with so many more consumers.”
7) Holding a loose identity fosters adaptability. Bret described a turning point at Facebook when Sheryl Sandberg told him, “You’re not holding your team to a high enough standard.” He realized he was “conforming the role to my identity” rather than adapting himself. “I woke up every morning and thought rather than come at this with a strong sense of my identity… I’m going to think what is the most impactful thing I can do today.” That flexible mindset allowed him to jump between engineering, product, CEO roles, and board chairs.
8) Clarity on a job’s value and broad adoption of AI are vital to survive disruption. He referenced the “Jobs to Be Done” framework by Clayton Christensen: “Companies confuse how they deliver something with why people want it.” In a new AI world, leaders must refocus on the “value they provide” and accept that operational methods will change. He suggested all firms should “build a customer-facing AI agent” and “officially sanction” AI for employees, saying “the more your employees use AI, the more they’ll come to you with meaningful ways to improve the company.”
o1
Transcript