OpenClaw Bulgaria Meetup #4 🦞 Observability Instrumentation for AI Agents
Agents are only as trustworthy as your ability to inspect them. Meetup #4 zoomed in on the tools and habits that make agent systems explainable under real pressure.
Highlight
Featured talk by Iliyan Petkov
Highlight
Practical discussion of traces, metrics, logs, and evals
Highlight
Trust framed as observability, not magic
Highlight
Strong fit for builders running agents in production or near-production
What this meetup was about
The core question was simple: when an agent does something strange, can you reconstruct why? That pushed the conversation toward instrumentation instead of vibes.
Iliyan Petkov brought a cloud-native perspective to the room, connecting modern observability practices with the emerging reality of agent workflows and tool calls.
The session stayed grounded in practical trust. Reliable agents need visibility, not just better prompts.
How the evening was structured
Featured talk
Observability Instrumentation for AI Agents, presented by Iliyan Petkov, focused on making agent behavior inspectable and debuggable.
Q&A with builders
Questions centered on what data should be captured, how to reason about agent failures, and how to move from intuition to evidence.
Open floor
People stayed to swap notes, compare stacks, and connect observability ideas to their own agent systems.
Resources
Links from Meetup #4
Useful follow-up material connected to this event.
Next step
See the wider OpenClaw Bulgaria event series
Each meetup adds another layer to the builder network around OpenClaw Bulgaria. Browse the full events page or check what is coming next in Sofia.