FDA/EMA regulatory intelligence · source-linked precedent
Regulatory Intelligence is ChatQP's advisory capability for regulatory-affairs and CMC (Chemistry, Manufacturing and Controls) teams. It reads the public assessment reports of the European Medicines Agency (EMA) and FDA review documents, and turns them into precedent you can trace to the exact source page.
It is integrated into ChatQP and runs locally. It is advisory by design: it surfaces what the approved record shows and points to the source — it does not predict acceptance, and it is walled off from the hard rules so it can never change a certification decision. Currently in pilot.
What it answers
Plain questions a regulatory or CMC team asks before a submission or a variation — answered from the approved EU and US public record, not from a model's general knowledge.
For a given molecule type: the control strategies, analytical techniques, storage and stability approaches, and variation history on record across authorised EU products. Authorised and refused dossiers are kept separate; a refusal is shown as assessment discussion, never as accepted precedent.
Which control strategies and analytical techniques actually appear in approved EU dossiers for a given quality attribute (CQA — critical quality attribute), with the exact source page behind each fact.
How a given type of CMC change has been handled historically — the variation history and the precedent cases behind it, with related FDA review evidence where it exists. Context for your judgement, not a prediction of the next decision.
How it reads the record
It reads the public EMA and FDA record and organises it into source-linked, review-ready precedent, traced to the original document. It works only from the public record — no general-knowledge inference, no invented facts — and assembles what is already there into a single working view across both registries, so a regulatory or CMC team can scope a submission from the evidence rather than start from a blank page.
Why it's trustworthy
The method is the point. Regulatory Intelligence assembles and organises facts that already exist in the public record; it does not generate them. Everything is built to be traced back to source and audited.
Every EMA fact links to the exact public assessment PDF and the page it came from; FDA-side evidence links to the review it was drawn from. You can open the original, confirm it yourself, and cite it in your own work — EMA gives exact-page traceability, FDA-side facts are advisory (see below).
It organises curated, source-linked evidence; it does not invent it. Where extraction from a document is thin or uncertain, it is flagged rather than filled in.
It is advisory by design, walled off from ChatQP's hard GMP rules and the certification verdict. It can inform a decision; it can never make or change one.
Numeric acceptance criteria and limits are deliberately excluded, so a report can never be mistaken for a specification. FDA-side facts are advisory and not human-validated — each is shown with its source to verify against.
What it is not
Status
Regulatory Intelligence is integrated into ChatQP as an advisory capability and is being piloted with a small group of evaluators. It draws on a curated, source-linked corpus of the public EMA and FDA record, surfaced as a deliberately small pilot. Benchmarks for this advisory work are not yet published; the batch-release work has its own benchmark outputs. If precedent intelligence is relevant to your regulatory or CMC work, we would like to talk.
ChatQP supports judgement. It does not replace the QP — and Regulatory Intelligence does not replace regulatory-affairs assessment. It points to the public record so people can decide with the evidence in front of them.