cleandataengineering.aiDrug-Safety Agent on Databricks, drawn simply
You are here: Checks

Drug-Safety Agent on Databricks · Checks

Why you can trust a number

A model can write fluent, confident text that is wrong. So nothing the model says reaches a person until deterministic code has looked at it. There are four places a check sits, in the order a request meets them.

WhereCheckFails howPage
Before the modelRefusal and hand-off rulesA fixed reply; the model is never calledwhat it will not do
Around the loopStep, cost and size limitsThe loop stops; a person is toldThe limits
On the draft's shapeOutput contract: required fields, right types, one repair turnOne repair turn; then a hand-offOutput contract
On the draft's factsGrounding: every number equals a tool resultNot shown as a finished draft; a hand-off with the mismatchesGrounding
On the data the model readsLabel text is data, never instructionsStanding instruction plus a planted-text scenarioLabel text is data

Why deterministic. A check that is itself a model is one more thing to trust. These are a few lines of code each. They give the same verdict every time, and each can be tested with a table of examples.

Look inside