Manual screening of medical literature is one of the most time-consuming parts of pharmacovigilance — and one of the least valuable uses of a PV specialist’s time. With this release, DrugCard introduces the AI Agent: a built-in automation layer that categorizes literature for you, right inside your project.
[▶ Watch the demo video]
It lives inside your project
The AI Agent isn’t a separate tool you need to configure from scratch. It works directly with the sources and products you’ve already set up in your project, whether you’re monitoring local or global literature.
Choose how much you trust it
Configuration happens on a dedicated AI agent settings page, where you define exactly how the agent should process incoming literature. You can set this up separately for local and global monitoring, and the agent works across three levels of automation:
- Level 1 screens for pharmacovigilance relevance and discards anything that isn’t PV-relevant.
- Level 2 goes further, also discarding articles that aren’t relevant to your product — and for anything safety relevant, it suggests a categorization for your specialist to review.
- Level 3 takes it all the way: it automatically applies the suggested Safety Case and Safety Information categories.
Each level always includes everything from the level before it, so you can start conservative and move toward full automation as your team gains confidence in the agent’s output.
You also set a schedule — daily, weekly, or none, if you’d rather trigger runs manually.
Full visibility into every run
Once launched, the agent’s progress is fully transparent. A live dashboard shows overall progress and a real-time breakdown of results: how many articles were categorized as safety cases, safety information, or discarded, along with anything that failed to process or was skipped.
The dashboard also shows the time saved on each run, a list of the latest categorized articles by category, and run details — when it started and when it finished — so your team can always see exactly what the agent did.
Built for regulatory traceability
Every run is recorded in the audit trail, giving your team complete traceability for every categorization decision the AI Agent makes — the same standard of transparency you’d expect from manual screening, without the manual effort.
Why we built this
Literature monitoring generates a high volume of publications, and the vast majority of them turn out to be non-relevant. The AI Agent handles that first pass automatically, so your PV team can spend their time on genuine safety signals — not on filtering through routine, irrelevant articles.
Curious how the AI Agent would work for your monitoring setup? Contact us to see it in action.
