Sierra

When you process payments for thousands of banks and millions of people, support isn't a nice-to-have — it's infrastructure. Fiserv is replacing hold times and legacy IVRs with AI agents across voice and chat — introduced today, scaling to millions of conversations. We're proud to be building it with them.

Paychex was founded on knowing your name. Now their AI voice agent delivers that same local-office feel at scale, processing payroll 24/7 with a 93% positive customer effort score. Thanks to Paychex CX leader Elizabeth Roaldsen, CFA for sharing how her team built an AI agent that feels as personal as it is helpful. Link in comments ↓

Most benchmarks test either an agent's ability to find information, or take action. 𝜏-knowledge tests both at once. Even among the strongest models, there's significant room to grow — and we're excited to see how the frontier moves in the coming months. Read the full blog in the comments.

Announcing agentic performance benchmarking for Speech to Speech models on Artificial Analysis. We use 𝜏-Voice to measure tool calling and customer interaction voice agent capabilities in realistic customer service scenarios Even the strongest Speech to Speech (S2S) models today resolve only about half of realistic customer service scenarios end-to-end - a meaningful gap relative to frontier text-based agents on the same tasks. Voice channels introduce significant complexity: challenging accents, background noise, and packet loss, all while requiring fast responses, consistency across long multi-turn conversations, and reliable tool use. Performance also varies considerably by audio condition: in clean audio some models perform notably better, but realistic conditions continue to pose a challenge. Conversation duration also varies meaningfully across models, with implications for both customer experience and operational cost. About 𝜏-Voice: Our Agentic Performance benchmark is based on 𝜏-Voice (Ray, Dhandhania, Barres & Narasimhan, 2026), which extends 𝜏²-bench into the voice modality to evaluate S2S models on realistic customer service tasks. It measures multi-turn instruction following, support of a simulated customer through a complete interaction, and tool use against simulated customer service systems. The simulated user combines an LLM-driven decision model with realistic audio synthesis: diverse accents, background noise, and packet loss modelled on real network conditions. This complements our Big Bench Audio benchmark measuring intelligence and Conversational Dynamics (Full Duplex Bench subset) benchmark measuring conversational naturalness. Scores are the average of three independent pass@1 trials. We evaluate under realistic audio conditions using the 𝜏²-bench base task split across three domains: ➤ Airline (50 scenarios): e.g., changing a flight, rebooking under policy constraints ➤ Retail (114 scenarios): e.g., disputing a charge, processing a return ➤ Telecom (114 scenarios): e.g., resolving a billing issue, troubleshooting a service problem Task success is determined by deterministic checks against expected actions and final database state, consistent with the 𝜏²-bench evaluator. Key results: xAI's Grok Voice Think Fast 1.0 is the clear leader at 52.1%, averaging 5.6 minutes per conversation, the second-longest overall. OpenAI's GPT-Realtime-2 (High) (39.8%, 3.0 min) and GPT-Realtime-1.5 (38.8%, 4.8 min) follow, with Gemini 3.1 Flash Live Preview - High close behind at 37.7% (3.8 min). Speech to Speech is a fast evolving modality and we expect movement in rankings as we continue to add new models with these capabilities, and model robustness improves. See below for further detail ⬇️

“There’s a right and wrong time to use a dog pun.” BARK trained their AI agent to know the difference. Now it handles a quarter of customer conversations—without losing personality. 🐾 Read more ↓

It’s Day 2 of CCW UK Summit in London. If you’re here, why not stop by the Sierra booth!? We’ll be sharing live demos and real examples of how AI agents can help brands go beyond case resolution to create more personal, consistent, and meaningful customer experiences across channels. Stop by, meet the team, and let’s talk about the customer interactions you’re looking to improve.

SF – we're hosting a happy hour on May 20 to talk all things healthcare AI. Join us for an evening with Sierra's Healthcare Agent Development team, including Swathi P. Florian Dahlhausen and Ben Hadden, to learn about how we're transforming the end-to-end care journey. Space is limited! RSVP here: https://luma.com/nrwipa50

This winter, we welcomed Agent Engineering interns who dove right in with customers and made real impact – from launching new agents to elevating our voice capabilities. Some highlights: our interns helped build μ-Bench - an open multilingual transcription benchmark used across the industry, launched a multi-national hospitality agent, supported use cases across insurance, media, tech, and financial services, and so much more. If you want to contribute to the most important technological wave of our time, applications for Fall 2026 are now open! Come help us build what's next.

It's not just selling one thing and moving on. It's how do we build a long-term relationship so our clients are happy, they're successful, they're connecting with more of their customers? Reggie Marable on the BILLIONS podcast.