Honeycomb
Honeycomb is reorienting observability around AI agents — as both subject and operator
◆Recent moves
- 12d ago
Agent Timeline is GA!
⚡ SPARKAgent Timeline reaches GA, giving teams visibility into multi-agent, multi-trace AI workflows — LLM calls, tool invocations, tokens, and failures in one view. The clearest sign Honeycomb is building a product for observing AI systems, not just classic services.
View source ↗ - 1mo ago
Introducing the new Canvas
⚡ SPARKThe redesigned Canvas turns investigation into a shared surface for multiple humans and agents, with agents that act autonomously when alerts fire and executable knowledge that teaches them the system. This is the structural move the AI cadence has been building toward.
View source ↗ - 1mo ago
BubbleUp Insights
BubbleUp Insights auto-analyzes BubbleUp results to surface notable differences and likely root causes. Layers AI onto Honeycomb's signature anomaly-explanation feature rather than introducing a new surface.
View source ↗ - 1mo ago
Dark Mode
Dark mode arrives across the product, following system preference or a manual toggle. Cosmetic quality-of-life, framed around late-night incident work.
View source ↗ - 1mo ago
Ask Canvas: Now in Beta
Ask Canvas enters beta, letting users ask plain-language questions about the data on screen with context-aware chat. An incremental step toward the full Canvas investigation surface that later shipped.
View source ↗ - 2mo ago
Honeycomb Canvas Slack App: Now in Beta
The Canvas Slack app enters beta, letting teams query Honeycomb data from any channel without context-switching. Extends the agent surface into Slack, where incident response often actually happens.
View source ↗