Virtual Expert for Regulatory Assistance (VERA)

The Virtual Expert for Regulatory Assistance (VERA) is an AI-enabled support tool developed by the Digital CMC CERSI to accelerate the safe and consistent adoption of digital and AI technologies in pharmaceutical CMC.

It directly supports the CERSI mission to advance regulatory science by providing a practical bridge between complex global guidance and real-world implementation.

VERA acts as an intelligent assistant aligned to the DCRL framework, helping users:

  • Find and interpret regulatory expectations in a clear and structured way

  • Define context of use, risk and evidence requirements for digital tools

  • Support preparation for regulatory submission or inspection

  • Promote consistency and shared understanding across organisations and regions

‍ VERA uses the Open AI LLM to query a Retrieval-Augmented Generation (RAG) database containing documents only relevant to Digital CMC CERSI's scope. The database currently contains over 300 regulatory guidance documents (draft, discussion and published), standards and best-practice scientific documents (open-access elements of pay-wall materials only). The LLM is constrained to avoid hallucination and will report if the answer to a question is not contained in the RAG database. All answers include relevant supporting references.

‍A key feature of VERA is its intelligent filtering system, which enables users to refine searches by region, document type, organisation, and year range. This metadata-driven approach ensures that users can quickly narrow results to the most relevant regulatory context for their specific needs.

The platform is designed to be continuously updated as new guidance documents and standards are published, ensuring that the knowledge base evolves alongside the regulatory landscape. It provides a text-based, conversational interface where users can ask natural language questions about regulatory expectations, best practices, and technical requirements related to digital CMC tools.

‍NOTE: As with any AI search tool, the outputs must be checked against source material and relevant experts before making any decisions that may impact and regulatory requirements, expectations or patient safety.

The VERA Knowledge Base (RAG Database)

At the core of VERA is a curated Retrieval-Augmented Generation (RAG) database, which ensures that responses are grounded in authoritative and up-to-date sources. This database brings together:

  • Global regulatory guidance (e.g. ICH quality guidelines, FDA, EMA, MHRA and other agency publications)

  • International standards and frameworks (e.g. ASME model credibility standards, ISO and related best practices)

  • Scientific literature and peer-reviewed research (covering modelling, digital CMC, AI/ML, and validation approaches)

  • Industry white papers and cross-sector reports (reflecting real-world implementation challenges and solutions)

  • CERSI-developed materials (including the DCRL framework, case studies, and practical guidance)

Why it matters

This structured evidence base ensures that VERA’s outputs are traceable, transparent and aligned with current regulatory thinking, supporting confidence for both technical experts and non-specialist users. The LLM is restricted to only providing responses from the RAG database and provides the reference(s) and relevant page number(s) when responding to all queries.

VERA exemplifies the CERSI approach: combining science, regulation and innovation to enable the responsible use of emerging technologies.

By making complex regulatory knowledge more accessible, queryable and actionable, it helps: accelerate medicines development, Improve global regulatory alignment, strengthen trust in digital and AI-enabled approaches  and safeguard product quality and patient safety.