OpenClaw AI Agent Observability Demo
In this demo, we explore the visibility challenges of OpenClaw agent workflows and how PuppyGraph helps make sense of agent traces using graph analytics. As agentic systems grow more complex, traditional dashboards fail to capture the whole story. PuppyGraph maps agent runs as a graph, making it easy to trace execution paths across tools, models, and subagents. Instead of stitching together joins and recursive queries, you get a clear, end-to-end view of how each run unfolded. This provides deeper insights into what your AI agents are actually doing once execution begins.
Tech stack
Queries in natural language
• Which exact tool calls led to this failure?
• Which subagents are spawned repeatedly across runs doing the same work?
• Which tools are being retried excessively across runs?
Want to try it yourself?
We've open-sourced the sample dataset, graph schema, and graph queries on GitHub, so you can recreate this demo in your own environment.
