DevFest Montréal 2026

← Back to the programme
4-4 · Under PressureEnglish

Gemini live: Architecting, building and pricing conversational experiences.

Category
AI & ML
Format
Session — 40 minutes
Audience level
Intermediate

Learn how to architect, build, and price, end-to-end conversational voice experiences using Gemini Live and Firebase with DealCalm.com as case study. We will dive deep into Gemini Live’s most powerful capabilities: sub-second barge-in interruptions, native Voice Activity Detection (VAD), live function calling, and low-latency bidirectional streaming, to show you how to build speech interfaces that feel genuinely human. Next, we will tackle the elephant in the room: API costs. We’ll break down real-world token burn rates, explore how context caching slashes streaming costs, calculate the true cost per voice minute, and review practical metering strategies to ensure your product scales profitably without burning through margins. Finally, you'll see how to build a profitable model of a micro-SaaS, taking into consideration the operational costs and customer lifetime value. Developers and founders alike will walk away with the practical blueprint to build next-generation voice AI products, paired with the financial clarity needed to price them sustainably.

Speakers

Felipe Lujan

5 years as Google Developer Expert (GDE) in Cloud and AI. Passionate about LLM-powered experiences.