Documentation

AI Ops

Compose insights from monitoring, incidents, browser, logs, traces, and deployments. Every claim is labeled Observed, Likely, or Unknown.

What it does

Incident / alert explanations from collected evidence
Root-cause ranking with honesty badges
Operator recommendations on Mission Control
Digests and NL-style search over workspace facts (compose)

Epistemic labels

Observed — directly measured. Likely — correlated hypothesis, not proven. Unknown — missing evidence. We never invent uptime or causation.

Plans

Requires Pro+ (ai_insights).

Limitations

Not a generative LLM product — no OpenAI/Anthropic key is required; composition over evidence only.
Quality equals the data you ingest — quiet surfaces stay quiet.

In the product

Mission Control AI block and /app AI surfaces when entitled.