How Bret Taylor’s new company is rethinking customer experience in the age of AI

Sierra is building autonomous AI agents that can help customers resolve issues and complete tasks without human help.
© 2024 TechCrunch. All rights reserved. For personal use only.

We’ve been hearing about the notion of customer experience forever, the idea that we could improve customer interactions with brands digitally. So far, the results have been mixed at best.

Sierra, the new startup from Bret Taylor and Clay Bavor, thinks that AI agents could be the next technology frontier, not unlike websites or mobile apps that came before them: essential digital assets for every company, and ones that could ultimately deliver on the promise of digital customer experience.

Whether or not that’s true, the two founders fundamentally see AI agents as a new technology category, providing an entirely new way for customers to interact with brands to improve their overall experience.

“Our thesis is really simple. We think that conversational AI will become the dominant form factor that people use to interact with brands, not just for the sort of current trends like customer service, but really for all aspects of the customer experience,” Taylor told TechCrunch.

What that means is that customers can enter free-form questions and requests into a search-style box, and the AI agent should be able to understand that request and take action by connecting to whatever transactional systems are required. That can be tasks like looking up an order in an order management system or rescheduling a delivery in a scheduling system, as a couple of examples.

Taylor and Bavor acknowledge that it’s not always easy to connect to these systems, especially if they’re older. But most of the CIOs they’ve talked to have indicated that they’ve built APIs that connect to these older systems, making it much easier for Sierra to communicate with them.

Regardless, Taylor and Bavor recognize that there are some serious challenges and risks when it comes to humans interacting with these AI agents. “When you put an AI in front of customers, the value is a lot higher obviously, but the risks are a lot higher, too, with brand misrepresentation and hallucination — all the technical problems that are candidly the hardest problems in AI,” Taylor said.

These are not minor issues, particularly the hallucination problem, where large language models sometimes make up things when they don’t know how to answer a prompt. That could be potentially devastating to a brand’s reputation, depending on the nature of the answer.

While no company has solved hallucinations yet — and potentially never will — Sierra is working to mitigate the problem (but then, isn’t everyone?). The company’s software is based on the idea of autonomous agents. “What that means in practice is that there’s not a single model producing a response from a Sierra agent.” In fact, Taylor says, it sometimes involves as many as seven models, including one they have dubbed “the supervisor” that monitors answer quality, and if it deems the answer questionable, it sends it back for reevaluation. Taylor acknowledges that handling hallucinations is an ongoing research problem for the industry.

As though that weren’t enough to worry about, when it comes to handling customer data in an automated fashion, there are a whole host of regulatory and data privacy issues to deal with. But Taylor and Bavor say that their agents are designed to handle that as well.

Taylor believes that AI is fundamentally different from software as we’ve known it over the last 30 years, and it requires an educational component to help customers understand the power and the pitfalls. “So part of our go-to-market motion is both mitigating these risks [and] teaching our customers about how this new type of software works,” he said.

But the flip side of that risk is that it represents a huge opportunity for the company. “Anytime there is a sea change in technology, it opens a window of opportunity for smaller companies to explore that open space and really take some risks and try some new things,” Bavor said.

This new wave of AI will generate at least five to 10 meaningfully new independent enterprise software companies, Taylor said, not unlike when cloud and mobile came along. “There’s an opportunity for a new technology model. There’s no market leader in conversational AI right now because it’s new. It’s a year old, if that, and so, everyone’s figuring this out in real time,” he said.

Taylor, who is also board chair at OpenAI, doesn’t see the two companies competing or any conflict between the two, although one could certainly argue that they do. “We don’t see OpenAI as competitive, and I will obviously recuse myself if there is ever a potential conflict,” he said.

The founders also think a new platform should have a new approach to pricing, and they have designed an entirely new pricing model based on outcomes. Instead of tiered subscription fees or usage-based pricing we’ve seen with other software companies, they want customers to pay only for outcomes, when a problem is resolved.

“We think outcome-based pricing is the future of software. I think with AI we finally have technology that isn’t just making us more productive but actually doing the job. It’s actually finishing the job,” Taylor said. And that’s the point where they intend to charge the customer. The mechanics, however, are still being worked out with early customers.

For all that, and even factoring in the experience of the two founders, Brent Leary, founder and principal analyst at CRM Essentials, thinks the usual incumbents like Taylor’s former company, Salesforce, are going to be difficult to compete with.

“I mean [Taylor] is incredibly intelligent and capable, there’s no doubting that,” Leary said. “But with Salesforce there’s a lot of institutional experience and skills and other resources that a startup doesn’t have, even if it’s headed by someone like Bret. And these huge companies are throwing all of their R&D investments and restructuring their whole operations already around the opportunities they’re seeing with AI.”

To be clear, Sierra is well capitalized, although certainly not at the level of a company like Salesforce. The pedigree of Taylor and Bavor combined with the potential market they are going after is attracting big investment with the company scoring $110 million already with $25 million from Benchmark, and an additional $85 million from Sequoia. That is an extraordinary amount of money for an early-stage company — but these are not your typical founders.

Sequoia Capital partner Ravi Gupta, who is leading his firm’s investment in Sierra, says beyond the background of the two founders, the firm was impressed by the technology and its potential. “I think seeing it in action is the thing that was remarkable, and I think it really captured our imagination of what future customer interactions can be,” he said, adding that it wasn’t a hard decision for him to write a check.

Sierra clearly sees a big opportunity to transform customer experience with AI, but many obstacles stand in the way of success. If the founders can find a way to adequately address the pitfalls of free-form, AI-driven, automated customer service agents, while staving off established enterprise competitors, it could be a successful startup, but like everything else involving AI, it still has to prove that it can do that — and do it consistently and at scale.

 


Comments

Leave a Reply

Your email address will not be published. Required fields are marked *