If you’ve built multi-turn or orchestrated AI agents in production, you already know traditional logging just doesn’t cut it. When an agent stalls, is it a latency spike in your vector database, an API gateway bottleneck, or a slow LLM response? Finding out usually means digging through siloed logs or troubleshooting your bleeding edge applications running on sometimes less than ideal tools.
To help you, we rolled out an update to the Solace Agent Mesh (Enterprise Edition) focused on day-2 operations. Read about it here - OpenTelemetry Observability in Solace Agent Mesh | Solace
Here is the short-version breakdown of what you can do with it right now:
- Zero-Friction Observability: We’ve exposed a rich set of real-time metrics and telemetry via a Prometheus-compatible endpoint.
- Real-Time Token Economics: No more surprise bills. Break down prompt/completion usage by model and task.
This announcement is part of a new, fancy changelog we’ve officially stood up where we’ll be dropping quick overviews with links to deep dives, codelabs, docs, and more on feature updates exactly like this moving forward. It is RSS-friendly and you can watch announcement in the community for the latest - What’s New in Solace - Latest streaming & AI Features | Solace.