Observed arrival · 2026-09-07
Mirador, an instrument for explaining AI-agent behavior
Mirador is a self-hosted observability console for AI agents and workflows, tracking health, compliance, costs, latency, and the reasons behind changes.
Field notes
The public page previews a v1 console rather than exposing a working dashboard. Its sample workflow data connects operational monitoring with governance: a marketing-review run raises a finding about an external model and missing approval, while quote generation retains personal data beyond a configured period. The interface also traces provider latency, spend by flow and model, individual run costs, and output quality across historical runs. Mirador states that deployment is self-hosted and does not require an SDK.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue