Multimodal AI agents in production
AI agents are everywhere. Production-grade AI is harder.
Multimodal AI agents in production.
On October 1st, Quicksort is hosting an evening in Paris dedicated to the engineering reality behind multimodal retrieval, Voice AI and agentic systems.
Powered by The AI Collective, the evening is an opportunity for people building and deploying AI to share technical experience, challenge ideas and compare what actually works in practice.
Two technical talks, open Q&As, and plenty of time to connect with other builders.
The talks

Docling vs ColPali: multimodal retrieval in production
A practical comparison of two approaches to multimodal retrieval, exploring how they handle real world documents such as PDFs, tables and images, and the engineering trade offs that emerge when moving from prototype to production.
Format: 20 to 30 min talk + 15 min Q&A

Voice AI and agentic AI: enterprises need agents, not longer prompts
What happens when AI moves beyond answering prompts and starts taking action? Jermiah will explore Voice AI and agentic systems, and what it takes to build AI that can operate within real enterprise workflows.
Format: 15 to 20 min talk + 10 min Q&A
What to expect
- Technical talks from engineers building these systems
- Real world production challenges and trade offs
- Open Q&A and discussion
- A room full of AI engineers, technical leaders and builders
- Food, drinks and time to connect
No generic AI intro. No sales pitch. Just technical content, practical experience and good conversations.
Schedule
Doors open
Food, drinks, and networking with fellow builders.
Welcome & introduction
Opening remarks and a quick introduction to Quicksort and The AI Collective.
Issa Hammoud · Multimodal retrieval in production
A technical look at Docling vs ColPali, what they enable for multimodal retrieval, and the trade-offs that matter in production. 20 to 30 minutes + 15 min Q&A.
Jermiah Jerome · Voice AI and agentic AI
Exploring what changes when AI moves from answering prompts to taking action in real-world workflows. 15 to 20 minutes + 10 min Q&A.
Open networking
Food, drinks, and conversations with engineers, founders, and fellow AI builders.
Powered by The AI Collective
The idea is simple: get people who are actually building AI in the same room, share what they've learned, and have the conversations that don't always happen on stage.
The AI Collective is a global non-profit building the human layer for the AI era. It unites 200,000+ leaders, builders, and stakeholders across 100+ forums worldwide to democratize the frontier, build trust, and coordinate how society navigates the rapid acceleration of technological progress.