Last October, we packed one day in San Francisco for the first-ever voice AI summit. It was one of our favorite days of the year, and one that changed the trajectory of Vapi and signaled the excitement around voice AI more broadly. A lot has happened since. The technology has moved fast, new frontier models seem to land every week, and more companies are putting voice agents into production. So we're doing it again, bigger: two days at Fort Mason, two tracks, 1,200+ builders and operators, and a speaker list that keeps getting better every week. You'll hear from the people shaping voice AI, the brands running voice agents at scale, and the team building the frontier at Vapi.
Here's everything we've announced so far.
Voice moved fast this year. So did the conversation.
A year ago, the question at VapiCon was whether a voice agent could hold a real call. That one's been answered. The questions now are harder and more interesting. Speech-native models are shipping on real phone lines. Latency that used to be a demo trick is the baseline. Teams that ran one agent last fall are running fleets in production, and they're showing up with real ROI and business results, not a demo and some slideware.
That's what VapiCon 2026 is about. Two days on what's working, what broke, and what's coming next, from the people building this frontier and the people running it in production.
The lineup so far
So far, we have speakers confirmed from a wide range of companies. They work on different applications and solve different problems in voice, but all are at the forefront of shaping our industry's future.
That starts with the people building the models and infrastructure: the teams behind how voice agents hear, speak, and connect, from speech recognition and voice synthesis to the real-time audio and telephony layers underneath.
- Scott Stephenson, Co-Founder & CEO, Deepgram
- Klemen Simonic, CEO, Soniox
- Neil Zeghidour, CEO, Gradium
- Brandon Yang, Co-Founder, Cartesia
- Kwindla Hultman Kramer, Co-Founder & CEO, Daily
- Andy O'Dower, VP, Field CTO, Twilio
It also includes the people shaping how developers build with AI: builders who've changed how software gets written and shipped, and who are now thinking hard about what agents do to that process.
- Harrison Chase, Co-Founder & CEO, LangChain
- Guillermo Rauch, CEO, Vercel
We're also excited to hear from the people making sure it actually works. Evals and observability are where many production teams are stuck right now, and these are two of the leading voices on the problem.
- Brooke Hopkins, Founder & CEO, Coval
- Laurie Voss, Head of Developer Relations, Arize AI
And of course, you'll hear from the team at Vapi. Jordan opens with the product and company update, including new releases and where the platform goes next. Nikhil shares where he sees voice heading. Vapi engineers will walk through what they've shipped, what they're building now, and how to get the most out of the platform.
- Jordan Dearsley, Founder & CEO, Vapi
- Nikhil Gupta, Founder & CTO, Vapi
More names are coming, including customers, builders, and leaders who've deployed voice AI at scale.
The themes we'll be digging into
We'll announce the full agenda shortly. What we can tell you is what's on our minds, and what every session ties back to.
Where voice AI actually is right now. Real call volumes, real failure modes, and the ROI teams are seeing when they deploy voice at scale. Expect rapid-fire pitches from developers running voice in production, plus a look at the "human" jobs voice AI is quietly matching, from coaching to collections, where even the end users were shocked they were talking to AI.
The architecture debates that matter. Speech-to-speech models hear tone and hesitation and respond fast, while a cascaded pipeline lets you control every layer of the stack. We'll also discuss the tradeoffs between open weights, open source, and closed, and the types of evaluation, testing, and observability that keep your agents on track.
Trust and reliability at scale. Getting an agent to work once is easy. Getting it to work on the ten-thousandth call, with interruptions, accents, background noise, and a caller who changes their mind mid-sentence, is a different problem. Sessions on sub-second time to first byte, fallback logic, human handoff, and why synthetic callers pass tests real humans fail.
Voice in the real world. Healthcare, contact centers, regulated industries, and a few deployments you wouldn't expect. This includes contact center leaders asking whether the metrics that measured humans are still relevant, and what successful deployments actually look like.
What Else. Multilingual agents that follow a caller as they switch languages mid-sentence. Grounding agents in governed enterprise data. And panels where a consumer builder, a research lab, and an enterprise platform argue about what ships in the next 12 months.
Why you should be in the room
Two tracks run side by side. The builder track is for engineers: workshops, demos, releases, roadmap. The operator track is for product, business, and ops leaders: case studies, panels, and real results.
If you're a CTO, an engineering or product lead, or the person driving your company's voice strategy, you'll want to see this. Bring the engineer who owns your agent and the operator who owns the number it's supposed to move.
See you at Fort Mason
November 11 and 12, 2026. Festival Pavilion at Fort Mason, San Francisco. Tickets are live, and the lineup is still growing.
Get your ticket at vapicon.ai
Speaker lineup as of September 2026. More speakers and the full agenda are coming soon. Check back regularly at vapicon.ai for updates.
