Agent Night: Live web data, memory, identity & actions
An evening for people already building agents. If you've shipped one, you know the demo isn't the hard part. In production, agents struggle to read the live web, forget context between sessions, and hit auth and permission problems the moment they act inside a real app.
This night covers three layers that address that - reading the web, persistent memory, and secure actions with the teams demoing each one. Three short talks, a Q&A, then time to talk to people working on the same problems.
Schedule
5:30 - Doors
6:00 — Welcome
6:15 — Bright Data: reading the live web with structured data
6:45 — Cognee: giving agents memory that persists across sessions
7:15 — Scalekit: secure, permissioned actions on behalf of users
7:45 — Q&A + networking, food and drinks
What You'll Leave Knowing
Why data quality, not model size, is often the bottleneck for agents in production
Where graph-based memory beats plain RAG, and when you actually need it
How agents can act on behalf of a user - delegated auth, scoped permissions, audit trails - instead of shared service accounts
How teams are wiring data, memory, and auth together in real production agents today
Who Should Come
People already building and shipping with agents: AI engineers, developers using tools like Claude Code and Cursor, platform engineers, technical founders, and GTM folks working hands-on in the space (solutions engineers, DevRel, technical PMs, product marketers close to the product).
This isn't an intro to AI agents. It's for people already past that.
Admission is free and approval-based because space is limited. Food and drinks provided.
Speakers
Swetha Kolli — Bright Data, Solutions Architect
Dave Nielsen — Founding Head of Developer Relations, Cognee
Tamilselvi Ramasamy — Product Marketing Manager, Scalekit
When: [DATE] | 5:30 PM – 8:00 PM PT
Where: Bright Data Web Data Loft, 625 2nd St, San Francisco, CA