What Honeycomb changed

• Added new platform features: AI Ecosystem, Agent Fleet Performance, LLM Cost Tracking, and Agent Conversations.
• Updated navigation to include "Canvas Connectors" and "Agent Skills."
• Replaced general AI blog posts with specific product announcements about closing the gap in "shipping agents."
• Renamed "MCP" to "Honeycomb MCP" to emphasize their proprietary integration.
• Moved "Anomaly Detection" from a blog link to a core platform page.

Honeycomb is shifting its focus from general LLM monitoring to specific tools for managing "AI Agent fleets." This targets companies building complex, multi-agent systems that need to track costs and conversations across many agents at once.

• Added a new technical blog post titled "How Adaptive Tail Sampling Works in the OpenTelemetry Collector."
• Removed an older tutorial post titled "Relational Query Superpowers."

Honeycomb is shifting its educational focus toward technical deep dives on OpenTelemetry data management. This helps them position their platform as the best way to handle complex telemetry pipelines.

• Added "Refinery as a Service" to the Core Observability product list in the main navigation.
• Updated the "Keep learning" blog section with a new post about how their "Canvas" tool helps AI agent development.
• Removed an older blog post about AI norms and values from the featured list.

Honeycomb is expanding its managed offerings by turning its "Refinery" sampling proxy into a hosted service. This reduces the work for teams who want to manage large amounts of data without running their own proxy servers.