An AI investment dashboard should not only show an answer. It should show why the answer changed.

That is why Invest AI added decision observability: decision events, macro radar factors, cache freshness, and worker-owned explanations that can be traced back to structured fields.

The problem

Without observability, a user sees:

Protect Capital
13 / 100

But they do not know whether that came from volatility, macro stress, weak breadth, stale data, or engine disagreement.

The shape

Decision observability turns the payload into a timeline of causes:

• Consensus events.
• Macro radar events.
• Engine events.
• Signal candidates.
• Stale/degraded state.

These events are generated in the worker cache, not invented in the frontend.

The UX rule

The top of the product stays decision-first. Details are progressive:

1. Header and hero card say what to do.
2. Event feed explains what changed.
3. Macro radar shows the factor layer.
4. Dashboard lets users inspect the full state.

Takeaway

Observability is not only for operators. In decision products, observability is part of user trust.

Further reading: ADR-013 (docs/adr/013-v11-decision-observability.md).