Kong Konnect AI Observability

Kong is previewing advanced AI observability capabilities for Kong Konnect that treat an entire AI session, rather than an individual request, as the primary unit of analysis. The approach is intended for complex, multi-turn agentic workflows involving multiple model calls, tools, MCP servers, policies, retries, and agents, where request-level telemetry may not reveal why a system’s behavior deteriorates. Session-level tracing connects these operations so developers can inspect inputs, outputs, token usage, cost, latency, failures, guardrail execution, and model or tool activity across a complete interaction. Kong also aims to provide unified analysis of performance metrics such as model success rates, time to first token, cache hit rates, outliers, and cost per request or session across AI infrastructure. A forthcoming Konnect Debugger is planned to supplement post-hoc tracing with live investigation capabilities, supporting a workflow of observing, tracing, investigating, and debugging AI behavior in context.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.