Milliseconds to Magic: Real‑Time Workflows using the Gemini Live API and Pipecat
summary
Demonstration shows Gemini Live API GA powered by Gemini 2.5 Flash combined with the Pipecat framework to run real‑time multimodal voice agents handling session management, turn detection, async function calls and telephony integration. The demo builds a list-and-UI app that mostly works — it updates grocery/reading/work lists and renders an animated “hello world” screen — but exhibits segmentation faults in intent resolution, name‑spelling errors and turn‑detection variability, highlighting where engineering still adds the “magic.”
AI-written summary · about
chapters (5)
- 0:00Voice as natural interface
- 2:11Voice AI magic and stack
- 6:44Demo introduction
- 16:01Demo wrap-up and Q&A
- 19:12Future of voice interfaces
related talks
56mBuilding Multimodal Apps with Gemini 2 0 and Gemini SDK
18mIntroduction to Gemini APIs and AI Studio
01h 01mBuilding Multimodal Apps with Gemini 2.0 and Gemini SDK
01h 48mBuilding Conversational Agents
01h 16mI'm in my Gemini Era!
21mBuilding at the Speed of Thought
02mUsing Gemini AI workflows for productivity
37mBuild agents with Gemini API