Agents After Hours: Securing What You Ship
A side event for AI Engineer Europe. Lightning talks and conversation about what it takes to ship agents that hold up in production.
Hosted by Sondera.
Day 2 of AIE Europe covers harnesses, context engineering, evals, and observability. This is the evening version of that conversation, focused on the security and governance side. If you spent the day thinking about how to build agents, come spend the evening thinking about how to keep them from breaking things.
You don't need an AIE Europe ticket to attend. If you're building agents in London or passing through for the conference, you're welcome.
Drinks and food provided.
The evening
19:30 — Doors open, less than a five-minute walk from the conference venue. Food and drinks.
20:00 — Lightning talks
Matt Maisel, Co-founder & CTO, Sondera - Governing coding agents with policy enforcement What happens when you treat agent permissions as code instead of trusting the model to police itself. A look at deterministic controls for tool use, file access, and data flow in coding agents like Claude Code and Cursor.
Preeti Ravindra, Senior Research Fellow, Pivotal Research
21:15 — Open networking
22:30 — Close
Topics on the table
Policy enforcement for agent tool calls: deterministic vs. probabilistic, and what survives an audit
Data flow control: preventing sensitive information from leaking through agent actions
Threat modelling for multi-agent systems: prompt injection, privilege escalation, tool misuse — What the EU AI Act, ISO 42001, and NIST AI RMF actually mean for teams shipping agents today
About Sondera
Sondera builds the security control plane for AI agents — deterministic, action-layer governance so teams can ship agents without losing sleep.