Sierra's Bret Taylor puts a four-year clock on the AI phone agent shift

Why it matters

Taylor is moving the AI-agent debate from cost savings to consumer preference. If that shift holds, call-center software becomes a brand surface, not just back-office tooling.

Bret Taylor (@btaylor), the co-founder of Sierra, said consumers will prefer businesses that use AI phone agents over those that rely on traditional call centers within three to four years, putting a 2029 to 2030 deadline on one of the largest labor shifts in customer service.

Taylor made the prediction in a clip posted by TBPN on X on June 23, 2026. His example was not a bank fraud line or an airline cancellation desk. It was a restaurant reservation. "If I have to call a restaurant to make a reservation, I'm going to a different place," Taylor said, before pointing to OpenTable as the kind of software interface that already changed consumer behavior.

https://x.com/tbpn/status/2069458030530560281

That is the sharper part of Taylor's claim. Sierra is not just selling automation to CFOs looking to reduce call-center costs. Taylor is arguing that the presence of an AI agent will become a consumer-facing product feature, closer to online booking or mobile checkout than to an internal efficiency project. The bet is that the phone, long treated as the channel customers use only when self-service fails, becomes programmable enough that customers begin to punish companies that still make them wait on hold.

Taylor has reason to make that argument. Sierra, which he co-founded with Clay Bavor, sells AI agents for customer experience across voice and digital channels. Taylor's own career sits at the intersection of enterprise software and consumer interface shifts: Sierra's founder bio says he was most recently co-CEO of Salesforce, founded Quip, served as CTO of Facebook, co-created Google Maps at Google, and serves on OpenAI's board. Bavor, according to the same company bio, spent 18 years at Google and led Google Labs, Google's AR/VR work, Project Starline, Google Lens, and Google Workspace product and design.

The founder story matters because Sierra is not trying to look like a generic chatbot vendor. From its February 2024 launch, the company has pitched agents that can perform work, not just answer FAQs. Fortune reported at launch that Sierra had raised $110 million from Sequoia Capital and Benchmark and was already handling customer conversations for WeightWatchers, SiriusXM, Sonos, and OluKai, according to the company. Fortune also reported that Taylor and Bavor first worked together at Google and decided to start Sierra after meeting for lunch following Taylor's January 2023 departure from Salesforce.

Sierra has since raised at a pace that reflects how aggressively investors are underwriting the agent thesis. On May 4, 2026, Taylor and Bavor wrote in a Sierra blog post that the company was raising $950 million from new and existing investors, led by Tiger Global and GV, at a valuation of more than $15 billion. Sierra said the round gave it more than $1 billion to invest and claimed it was serving more than 40% of the Fortune 50, with agents on its platform powering billions of customer interactions.

Those are company-supplied numbers, but they show the scale of the market Sierra is trying to define. The company said in November 2025 that it had reached $100 million in annual recurring revenue seven quarters after launching in February 2024. In that same post, Sierra named customers including Deliveroo, Discord, Ramp, Rivian, SoFi, Sonos, Tubi, Wayfair, Next, ADT, Bissell, Safelite, Vans, Cigna, SiriusXM, and DIRECTV. It also said agents built on Sierra can work over phone, IVR, chat, WhatsApp, email, and ChatGPT, and that 50% of companies using Sierra had more than $1 billion in revenue while 20% had more than $10 billion.

Sierra's public customer page now lists a broad set of case studies across financial services, healthcare, media, retail, telecommunications, travel, and home services, including SoFi, GoFundMe, Airtable, Redfin, Chime, Ramp, CLEAR, Sonos, SiriusXM, ADT, Casper, Minted, WeightWatchers, and Brex. The site includes self-reported metrics such as an 80% resolution rate for Airtable, a 90% case resolution figure for Ramp, 34 million SiriusXM subscribers, and 2 million monthly customer inquiries for ADT.

Taylor's phone-agent prediction also reveals Sierra's real competitive wedge. The obvious enterprise buyers want lower service costs, shorter wait times, and fewer escalations. But Sierra is trying to make the buyer believe that AI agents will affect revenue, retention, and brand preference. That is why Taylor's restaurant example matters. A reservation is a low-stakes transaction where the customer has alternatives. If consumers begin to apply that same impatience to insurance claims, banking, subscriptions, healthcare scheduling, or travel changes, then call centers become a liability rather than merely an expense line.

The hard part is that voice agents carry less tolerance for failure than text chatbots. A bad chatbot can be abandoned. A bad phone agent can trap a customer in a loop, misunderstand an urgent problem, or break trust at exactly the moment the customer needs help. Sierra has tried to answer that concern by emphasizing enterprise guardrails, regulated-industry use cases, and agents that can escalate to humans. But the category's adoption will depend on whether companies can make AI phone agents feel faster and more reliable without making customers feel blocked from a person.

That is the test behind Taylor's timeline. If he is right, AI phone agents will not be sold only as software that deflects tickets. They will become part of how a consumer judges whether a business is modern enough to deal with. If he is wrong, the agent market still grows, but as a cost-saving layer inside the contact center rather than the new front door for customer relationships.