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

Drug-Safety Agent on Databricks · Status

What is done, what is not, and the order

Of the 35 steps in one investigation, 28 are built, 6 are partly built and 1 are not built. The order below closes the most gaps first.

#Next piece of workWhat it closes
1One test per owed harness capability, or an accepted gap with an owner and a dateThe harness gate, now 2 of 76
2The open-question case: what should I look at for this drug?, built and testedThe one place a model is genuinely needed; today's agent is mostly a deterministic core
3Move the store to a Delta table; put a login in front of the desk; notify a reviewerThe two stand-ins in the desk
4Write three hand-off reasons, the freshness window and the erasure rule into the specificationFindings from building and running it
5Wire the agent to the live Unity Catalog functions; retire the Python copyOne definition of each tool
6Run the 20-pair evaluation on the Databricks model endpointEvidence for the 'built on Databricks AI' claim
7Stronger scenario checks that also test what the reply saidScenarios that cannot pass by only calling a tool
8MLflow tracing and the shared evaluation library; retire the old first-draft loopObservability; dead code

What the demo can honestly claim. A working, tested, rule-bound drafting assistant over public data. A weekly job picks the pairs worth a look; drafts wait in a queue; a named person decides under their own name, and a decision is final; a case the agent cannot finish goes to a person with a reference. Every number in a draft is read from the data. What it cannot yet claim: that it runs on Databricks end to end, that the desk is secured, that the store is the production one, or that the model's open-question ability has been tested.