Academic Research Services (ARS) is excited to announce the launch of the Information Commons (IC) Documentation Copilot—a new AI assistant that helps UCSF researchers and IC data users find, understand, and apply IC documentation faster.  

The Copilot is designed for the reality of IC work: multiple data assets (e.g., DEID CDW, OMOP), multiple tools (e.g., PatientExploreR, OHDSI ATLAS), multiple secure environments (e.g., RAE, IC FAC, AWS Secure), and lots of helpful—but often fragmented—documentation. 

What the Copilot does 

The IC Documentation Copilot is built to answer real researcher questions by retrieving IC sources first (documentation + precedent), then composing a practical, researcher-friendly answer with citations you can click to verify. 

You can use it for: 

• IC ecosystem navigation & access questions (e.g., which environment/tool to use and how to get started)  

• Conceptual data questions (data provenance, clinical workflow context, cohort logic, pitfalls) 

• Applied “how-to” guidance, including workflows and example query patterns when helpful (with citations) 

What’s in the knowledge base (at launch) 

At release, the Copilot is grounded in a curated IC knowledge base including Slack Q&A threads, IC wiki content, and data dictionaries (with ongoing expansion). 

How to access 

1. Open Versa Chat 

2. Go to Assistants → select Information Commons Documentation Copilot  

3. Recommended model: GPT-4o  

No Versa Chat access? Email [email protected]. 

Tips for best results 

• Name the dataset in your question (e.g., “DEID CDW” vs “OMOP”).  

• For table/column lookups, ask it to confirm in the data dictionary (and cite the variable description). 

• Use citations to verify quickly.  

Support and feedback 

Your feedback helps us improve coverage and answer quality. After an answer, use the 👍/👎 feedback option and include a short comment on what worked or didn’t. 

Questions or help: 

• Slack: #ucsf-deid-cdw-omop 

• Email: [email protected] 

Acknowledgements  

This release was made possible through strong collaboration with the UCSF Versa team. We want to thank Versa Chat development team for partnering with ARS to build and tune the underlying vector databases, integrate IC knowledge bases, and implement the IC Documentation Copilot assistant within the Versa Chat platform.