Honeycomb, the observability startup formerly known as Hound Technology, today unveiled a suite of updates designed to give engineering teams deep visibility into AI agent activity in production. The new features include Agent Timeline, Canvas, and Canvas Skills, all of which require no proprietary SDKs or specialized frameworks.
"AI agents are now part of the engineering team," said co-founder and CEO Christine Yen. "But right now, most teams can't see what those agents are doing in production: which tools they called, what they decided, whether they made things better or worse."
Agent Timeline offers a single dashboard connecting every LLM call, agent handoff, and tool invocation, allowing engineers to trace activity, reconstruct decisions, and understand failures in real time. The revamped Canvas serves as a chat interface and autonomous agent, enabling plain English queries for system insights. Canvas Skills let teams teach AI agents routine debugging protocols as reusable playbooks, automating repetitive investigations. Auto-investigations allow Canvas to trigger automatically on alerts, running pre-built playbooks against anomalies, gathering data, testing hypotheses, and suggesting responses before engineers even reach their terminals.
Agent Timeline is now in early access, with general availability expected in a few weeks. The remaining updates are available immediately.