Bay Area Voice AI Night ｜StepFun X Seamate with PLAUD, Firework & Cresta
Voice AI Night is a Bay Area gathering for founders, builders, product leaders, and investors interested in how voice, multimodal AI, and real-time agents are changing the way people interact with products.
On July 30, StepFun and Seamate bring together people interested in how voice, multimodal AI, and real-time agents are changing the way people interact with products.
Join us for an evening of model insights, product discussion, outdoor dinner, and relaxed conversations with people building and exploring the next wave of AI products.
What's Happening
🎙️ Keynote : Voice models, from the model side
StepFun is a frontier AI lab building language, multimodal, and audio models. StepFun will share what it is seeing from the model side: where speech models are moving, what StepAudio 2.5 is designed for, and what better realtime voice intelligence could unlock for builders.
🧠 Founder panel: What It Takes to Ship Voice AI
Feifan Fan, Head of AI Research @ PLAUD
Vincent Yang, Co-founder and CEO @ Firework
A conversation on product lessons, application-layer opportunities, and what changes when AI can listen, speak, and interact in real time.
The panel will be primarily in Mandarin 普通话, and English-speaking attendees are very welcome.
🍽️ Outdoor buffet dinner + unhurried conversations
After the main session, the conversation continues outdoors over seafood, lamb and beef, with two full hours to meet and exchange notes with other founders and builders.
Agenda
5:00 PM ｜ Check-in & Welcome Drinks
5:20 PM ｜ Opening
5:30 PM ｜ Keynote + Founder Panel
7:00 PM ｜ Dinner Mixer, continuing from the main session
Who Should Come
People building, investing in, or seriously exploring the next wave of AI products.
You do not need to be deep in Voice AI already. We especially welcome application-layer builders and people curious about where voice and multimodal interaction are going next.
Registration is approval-based so we can keep the room relevant and useful.
🌟About StepAudio 2.5
StepAudio 2.5 is StepFun’s audio model family for speech understanding, speech synthesis and speech recognition.
It supports speech understanding with vocal cues, instruction-following speech generation, zero-shot voice cloning, and low-latency Chinese and English speech recognition for live captions, voice input, and meeting transcription.
Explore the official StepAudio 2.5 documentation →
Language
The keynote and panel will be primarily in Mandarin. English-speaking founders and builders are more than welcome to join. Networking and dinner will naturally be bilingual.
About the Hosts
Seamate (出海同学会) is a community for founders, builders, and investors exploring AI, startups, and global markets. With 15,000+ members across 10+ countries, we connect entrepreneurs across the U.S., China, and Southeast Asia and beyond through events, conversations, and collaborations.
StepFun is a leading foundation model company committed to the long-term pursuit of AGI, backed by a top-tier technical team. Its Step model family spans language, multimodal, and reasoning capabilities.
With a strategic focus on AI-powered intelligent terminals, StepFun has also built strong commercial traction across major consumer device categories, including smartphones and automotive. Its model capabilities have been integrated into large-scale device and in-car agent scenarios, bringing foundation model intelligence closer to everyday user experiences.