An interview with Bret Taylor and Clay Bavor, co-founders of conversational AI startup Sierra, on using several AI models at once, building AI agents, and more
Sierra hopes every company will use its empathetic AI chatbots for customer interactions. — I might have inadvertently … X: @wired , @fractaldoctor , and @artsjournalnews LinkedIn: Steven Levy and Mark Oehlert X: @wired : Sierra hopes every company will use its empathetic AI chatbots for customer interactions. https://www.wired.com/... Dr Stuart Woolley / @fractaldoctor : All we want, really, is just to talk to a human being who's not under pressure from their employer to sell us things, constrain us to a perfunctory time slot, or discourage us from unsubscribing. Removing the humanity from a conversation isn't the big win you think it is. @artsjournalnews : “In the future, a company's AI agent—basically the AI version of that company—will be just as important as their website,” says Taylor. “It's going to completely change the way companies exist digitally.” - Wired https://www.wired.com/... [image] LinkedIn: Steven Levy : I have known Bret Taylor and Clay Bavor for years, and hold them in high regard. So I was fascinated to hear about their startup, Sierra … Mark Oehlert : First, the pedigrees here if you don't know them: Taylor, former co-CEO of Salesforce and was a key developer of Google Maps in the aughts and Bavor headed Google's VR efforts. …
Context & Ripple Effects
Sierra entered the enterprise conversational-AI market with $110M in launch funding, led by founders whose experience spans Salesforce and Google. This interview clarifies the product thesis behind that launch: customer-facing agents that can draw on several models rather than a single model stack.
The company’s subsequent financing and valuation coverage indicates that investors treated this as a serious enterprise-agent bet, not merely a chatbot feature. The key question is whether Sierra can turn model choice and conversational design into a dependable customer-interaction product.
First-order effects
- Sierra is positioning its customer-interaction offering around multi-model orchestration and conversational agents, giving prospective enterprise customers an alternative to deploying one foundation model directly.
- The founders’ public framing puts agent quality—especially the ability to handle customer conversations—at the center of Sierra’s differentiation.
Second-order effects
- Enterprise AI vendors and customer-service platforms face pressure to show how their agents select, combine, or substitute models while preserving a coherent customer experience.
- Model providers become less visible to end customers when an agent vendor owns the interface and can route work across multiple underlying models.
Third-order effects
- If multi-model agent platforms prove reliable, competitive advantage in enterprise AI may shift from access to any one model toward workflow integration, evaluation, and control of the customer-facing interaction.
- The pattern points toward AI customer-support agents becoming a distinct enterprise software layer, though adoption will depend on whether companies trust them with consequential customer exchanges.
The trend: Enterprise AI is moving from standalone chat interfaces toward embedded, workflow-specific agents whose value comes from orchestration and operational fit rather than a single underlying model.