Context is King #2 – Building AI Agents on Real Data
Every AI team hits the same wall: your agent works on demos but breaks on real data. The context is wrong, incomplete, or doesn't scale.
The models are good enough. The hard part is everything around them – feeding agents the right context from large, complex data environments so they actually produce reliable results in production.
After a sold-out debut (200+ waitlisted), Context is King is back – this time in SF.
🎤 Speakers
Eno Reyes, Co-founder & CTO at Factory – Context compression for coding agents. They built an eval framework to measure how much context different compression strategies actually preserve across long-running agent sessions, and tested it against approaches from OpenAI and Anthropic.
Marko Bocevski, Head of Databases at Aiven – Why your agent is flying blind without data context. A demo showing the difference between operating with and without rich metadata, and how MCPs are turning static data catalogs into real-time context layers agents can reason with.
Severi Tikkala, CTO at ConfidentialMind – Context and RAG in practice. A walkthrough of a real RAG failure, fixed step-by-step using hybrid retrieval, metadata and scope filtering, and retrieval tracing – three practical "context knobs" used in production.
💻 Lightning talks / demos
**To be announced — Monday 23rd**
📅 Agenda
Doors open at 5:30 PM
5:30 - 6:00 PM: Check-in, Networking with light bites and drinks
6:00 - 7:30 PM: Presentations + Q&A
7:30 - 9:00 PM: Networking with food and drinks
Doors close at 9:00 PM
👋 About the organizers
Aiven is an open-source data infrastructure platform providing managed databases, streaming, and analytics — including PostgreSQL, Apache Kafka, OpenSearch, and ClickHouse — across all major clouds.
Flow AI builds the infrastructure layer that lets data and analytics SaaS companies deploy AI agents directly on their production data — with structured query execution, semantic data layer, and multi-tenant isolation.