Zeno Divergent

Guide · forecast reliability

Monitoring looks back. Zeno checks first.

Zeno learns when forecasts deserve trust before a decision depends on them. It complements ML observability rather than replacing it.

Try Zeno CheckThe model

observability

is my model still healthy?

forecast reliability

is this forecast trustworthy?

01 · difference

Two questions, side by side.

Use both: monitoring for the health of your own model, Zeno for the trust decision on each forecast.

Question answeredDid my model's inputs or accuracy drift?Should I trust this forecast, now?
WhenAfter errors pile upBefore the decision, per question
UnitOne model's metrics over timeA panel of forecasts on one question
UsesLogs, feature statistics, labelsForecasts, stated ranges, prior track records, shared mistakes
OutputAlerts and dashboardsKeep, adjust or review — with the reason and the evidence scope
When unsureRaises an alertReturns your forecast unchanged and says why

02 · limits

What Zeno does not claim.

It corrects only where later held-out tests support it, source by source.

  • not a drift detector
  • no free-text understanding
  • warnings only where proven
  • no trading advice
  • keeps your forecast when unsure