Why it matters
Tavus enables agents that see, hear, and respond in real time over video. Pairing real-time multimodal agents with rapid UI generation reveals cost structures well beyond simple text inference pipelines.
The tokenmaxxing angle
Real-time video agents process continuous audio/video alongside language inference, creating sustained token throughput unlike batch tasks. The event surfaces what latency-sensitive multimodal inference actually costs.
From the organizers
Venue: Tavus HQ, 35 Stillman St, San Francisco. Sponsors: Tavus (free API credits) and Lovable (credits + live URL). Teams of 3-5, lunch included, screening form required. June 27, 2026.