At Sierra AI Agents Are the Product and the Business Model

Before AI became the centerpiece of every enterprise pitch deck, Bret Taylor had already cycled through the industry’s defining platforms from building at Google and Facebook to running Salesforce as co-CEO. Now, as co-founder and CEO of Sierra, Taylor is focused on something narrower but more immediate: helping companies deploy customer-facing AI agents that actually work in high-stakes, high-volume environments.

Taylor frames the future of enterprise software around a clear shift that companies are moving away from fragmented digital tools toward unified AI agents that handle full customer interactions.“Every company will need their singular agent,” he said, “not plural—the one that has their brand at the top and represents the whole of their customer experience.” Taylor emphasized that successful vendors are those aligning directly with customer outcomes and operational needs.

Sierra’s clients, including ADT and SiriusXM, aren’t chasing novelty. They’re trying to lower support costs, manage surges in seasonal demand, and retain customers in competitive markets. That’s where Sierra’s agents come in: resolving subscription issues, triaging account problems, and helping users navigate services without friction. Taylor’s view is that the best AI companies will solve business problems precisely and repeatedly.

Engineering, Then Letting Go

Reflecting on his time as Facebook’s CTO, he described a moment of reckoning when then-COO Sheryl Sandberg pulled him aside. “She essentially said, ‘You need to hold your team to as high a standard as you hold yourself. You need to stop trying to do the work yourself.’”

At the time, Taylor admitted, he was trying to conform the job to his strengths rather than doing what the job required. “I was going into work every day and saying, ‘How do I conform this job to me?’ instead of thinking, ‘What’s the most important thing to do in this job today—even if it’s not something I find delightful or interesting.’” The lesson, he said, was not about efficiency, but self-awareness. “I got more joy from impact than from any individual act of doing the job.”

That same approach, prioritizing adaptability over comfort shapes his view of entrepreneurship. “At every stage of your company, what you need to be great at changes. Having the self-awareness to actually change what you spend your time on is one of the greatest challenges of being a founder.”

Building for Outcomes, Not Access

Much of Sierra’s product philosophy is reflected in how it prices. “We call it outcomes-based pricing,” Taylor said. For the median Sierra customer, that means the company gets paid when the AI agent successfully resolves an issue. If the case is escalated to a human, it’s free. “It felt like, let’s pay for a job well done. Salespeople get paid a commission—why not the AI?”

That model also forces Sierra to stay close to the customer’s business needs. Taylor pointed out that in traditional enterprise software, vendors often exit the process after the sale while systems integrators take over. “You end up with an active disassociation from outcomes.” Sierra’s pricing structure, in contrast, requires product and engineering teams to stay involved. “It really disrupts the way you build a software company.”

Taylor views this not just as a product choice, but a strategic moat. While incumbents remain bound by legacy revenue models, startups have more flexibility. “Most people in this room are not encumbered by a business model,” he said. “That’s one of the greatest advantages that startups have.”

What Agents Actually Do

The Sierra team sees agents not as support chatbots but as interfaces to the entire business. For Minted, a design-focused e-commerce brand, Sierra’s agent helps users upload address files for holiday cards, check order statuses, and match art to home décor styles—all while handling massive seasonal demand without human wait times. The agent also integrates with Minted’s backend systems, creating summaries of customer issues and surfacing trends in real time.

Taylor compared this to how websites evolved from static directories to full-service transaction platforms. “Companies have had many digital touchpoints—websites, apps, social profiles. But fast forward five years, and the majority of digital interactions will happen via an agent.”

That shift, he argued, requires rethinking software categories altogether. “I think this will manifest in the applied AI market as agents. This is the new form of software-as-a-service.”

He’s skeptical of horizontal platforms that aim to be everything to everyone. “Every industry has unique workflows. A telecommunications company doesn’t operate like a health insurer. If you’re solving a specialized problem, you can deliver value more quickly.”

Taylor also dismissed the idea that improved models alone would decide the winners. “The value is taking this next-generation technology and solving a high-cost business problem. If you do that, selling something valuable for an order of magnitude less than the current cost is really easy.”

Selling the Right Way

Despite his engineering roots, Taylor repeatedly returned to a recurring theme: great technology doesn’t sell itself. “One of the things I see a lot with entrepreneurs is they go into meetings and sell their product. But really good salespeople? They ask a lot of questions.”

His advice: research the company, show up informed, and frame the conversation around the customer’s problems—not your product’s features. “You find the intersection between what you do differently and where there’s actual business value.”

And when it comes to enterprise sales, understanding how your buyer makes decisions matters more than most founders realize. “Sometimes it’s easier for companies to prepay than to pay on demand. It depends on who you’re selling to, how they buy, who the gatekeepers are. That’s how you grow an enterprise software business.”

For Taylor, the AI moment echoes earlier cycles in tech. At FriendFeed, his social startup acquired by Facebook in 2009, the team built physical servers and managed their own co-location right down to the power cables. “We literally brought the site down once by tripping over a wire,” he said. “It was just a different era.”

Today, thanks to cloud infrastructure, open-source libraries, and virtualized financial tools, founders can focus more sharply on differentiation. But that also comes with a temptation to tinker with the parts of the business that don’t move the needle. “Every ounce of time you spend on something that should be a commodity is time you’re not spending selling or working on your core product.”

His advice is to build what will make your company stronger a year from now and not what feels gratifying this week.

“If you’re solving a real problem and making your customer successful, that’s where the truth is,” he said. “And that’s where your product should go.”