Clinical Research Document Intelligence for Cell & Gene Therapy

The business challenge

Clinical and regulatory writing across six CAR-T and gene therapy programs required synthesizing complex evidence while preserving scientific traceability. Clinical study report (CSR) drafting took 12 weeks, creating an opportunity to accelerate preparation without shifting accountability away from medical writers.

The architecture behind the outcome

I architected a GxP-validated GenAI copilot combining retrieval-augmented generation (RAG), multi-agent drafting, and a cell and gene therapy knowledge graph. The copilot writes and checks clinical and regulatory documents against source material, helping writers move from evidence gathering to informed review.

Delivery & decision ownership

• Ground document generation in retrieved evidence and domain relationships from the CGT knowledge graph.
• Use multi-agent drafting and checking to support clinical study reports and safety narratives.
• Embed human review into the operating model, with 98.6% of generated statements traceable to source documents.

Results that matter

The program delivered $3.8M in annual benefit across six CAR-T and gene therapy programs. Weekly adoption reached 82%, and the lead biologics license application (BLA) submission moved forward by approximately six weeks.

Source: Puspamitra Mishra’s Project Highlights Summary. Results are reported in that summary; the workflow illustrates the described solution and is not a production deployment diagram. Photography is illustrative.