.png&w=828&q=75)
RelAI is a messaging-first multi-agent networking platform built for AI Agent Olympics at Milan AI Week 2026. Instead of asking attendees to search directories or awkwardly approach strangers, RelAI assigns each person an AI representative that works on their behalf throughout the event. The experience starts in Telegram, where users answer a short onboarding flow about their role, interests, who they want to meet, and when they are available. Google Gemini transforms these answers into a structured networking profile and launches a coordinated agent workflow: scanning the attendee roster, ranking the strongest mutual fits, simulating short agent-to-agent conversations to test alignment and interest, finding overlapping availability, and proposing up to three high-quality meetings with clear reasons and suggested talking points. This is not a generic chatbot. RelAI implements a collaborative multi-agent system with distinct steps—profile extraction, matchmaking, representative dialogue, scheduling, and summarization—connected through a shared database and orchestrator. Users retain full control: every proposed match requires explicit approve or reject in Telegram before anything is confirmed. For organizers, judges, and power users, Mission Control is a web dashboard that visualizes the workflow in real time: a live graph of scanning, contacting, and negotiating states, an activity feed, and match cards with scores and transcripts. The stack is Next.js, Supabase Postgres, Gemini Flash and Pro, and deployment on Vultr, aligned with hackathon partners and tracks including Collaborative Systems, Agentic Workflows, Enterprise Utility, and Intelligent Reasoning. Enterprise use cases extend beyond conferences to sales roundtables, investor-founder matching, and internal offsites where quality introductions matter more than volume. RelAI demonstrates how autonomous agents can collaborate at scale while keeping humans firmly in the loop.
19 May 2026