Feedback

Research Finance Forum updates

16 October 2026      Joni Rhodes, Deputy Executive Director

AI in research


The October Research Finance Forum included a lively discussion on the potential opportunities and pitfalls arising from the use of AI in grant management and pre and post-award compliance. The discussion focused mainly on Copilot, though some institutional tools are being developed. Use of third party tools was seen as difficult due to integration with existing systems and privacy concerns. 

Thanks to Ashley Blake of RKE Companion, who we invited along due to his expertise in research management and AI tools, for the following useful summaries, resources, and information. Please note, these have note been reviewed or endorsed by BUFDG.  

Ways to use AI in research grants - resources

Prompting in Microsoft Copilot

Microsoft Copilot uses natural language prompts to help users draft, summarise, analyse and organise information. Effective prompting follows five principles: Context, Reference, Ask, Format and Test (CRAFT). Clear instructions, trusted sources and human verification improve accuracy and consistency.

AI prompt pack - document

Building Agents in Microsoft Copilot

Copilot agents are customised AI assistants designed to perform specific tasks, follow instructions and use approved organisational knowledge. Using Copilot Studio, organisations can build agents to answer questions, support workflows and automate repetitive activities.

Key difference: Prompting supports individual tasks. Agents deliver repeatable processes, consistent outputs and greater automation, subject to appropriate permissions and human oversight.

What Should an Organisational AI Policy Include?

An effective AI policy should establish clear rules for responsible, secure and transparent AI use, including:

  1. Approved AI tools: Which platforms staff can use and for what purposes.

  2. Data protection and confidentiality: Rules for handling personal data, unpublished research, intellectual property and commercially sensitive information.

  3. Human oversight: Who is responsible for reviewing, verifying and approving AI-generated outputs.

  4. Accuracy and transparency: Requirements for checking sources, identifying errors and declaring AI use where appropriate.

  5. Governance and accountability: Risk assessments, access permissions, audit trails, incident reporting and regular policy reviews.

The policy should align with UK GDPR, institutional research governance requirements and relevant international AI standards.

Five-Step Criteria for Responsible AI Prompting

The CRAFT framework: Context, Reference, Ask, Format, Test. These five steps should apply to all prompts to ensure outputs are accurate, consistent, verifiable and appropriate for use in UK research management.

  1. Context: Define the purpose. State the task, intended audience, relevant institutional requirements and the role AI should perform. 

  2. Reference: Provide trusted evidence. Supply relevant funder guidance, policies and source documents. Only share information permitted under your organisation's data protection and AI policies.

  3. Ask: Specify the task and boundaries. Explain exactly what AI must do. Require it to identify missing information, avoid assumptions and never invent facts, figures or requirements.

  4. Format: Define the expected output. Specify the structure, headings, word limits and level of detail. Require clear UK English and traceable references where appropriate.

  5. Test: Verify before use. Check all facts, calculations, eligibility criteria, dates and recommendations against authoritative sources. A qualified person must review and approve outputs before they inform decisions or submissions.

    Core principle: AI supports professional judgement. It does not replace accountability, evidence or human oversight.

    What Is ISO/IEC 42001?

    ISO/IEC 42001 is the international standard for establishing and maintaining an AI management system that helps organisations develop, deploy and use AI responsibly. It provides a framework for managing AI risks, accountability, transparency, governance and continuous improvement.

    See this ISO42001 presentation for more information. 

    Why Is the EU AI Act Important for UK Organisations?

    The EU AI Act can apply to UK universities developing or deploying AI systems where those systems, or their outputs, are used in the EU, making it relevant to international partnerships and collaborative research. It introduces risk-based obligations covering transparency, oversight and compliance, although specific scientific research exemptions apply, so universities must assess which activities fall within its scope.

    Discussion summary; While interest in AI was high, the consensus was that AI currently works best as a productivity and administrative support tool, with professional judgement, institutional guidance and human oversight remaining essential for compliance, funding and research management decisions.

    On to other updates or requests from the meeting...

    Innovate UK updates

    Attention was drawn to recent updates arising from the quarterly meeting with Innovate UK, and thanks were given to BUFDG's Innovate 'champions' for their ongoing involvement in these discussions:

    Lisa Weston - Cranfield University
    Jo Johnson and Darren Storey – Newcastle University
    Olvi Luir – University of Warwick
    Keri Dunning – University of Leeds
    Charlie Greene – UCL
    Lucy Haward – University of Southampton

    Innovate UK have asked for a set of FAQs e.g. what would be most useful to hear from them in relation to evidence requirements, and what do you struggle with most in the portal? If we can get a set of questions or navigation concerns from BUFDG members relating to IFS issues, we can arrange a session around this. Please email Joni with suggestions and we will collate and theme. 

    Elsewhere, we are setting up a call to discuss Innovate funding and VAT implications on 5 November. If you, or members of your tax team, would like to join the call please email Andrea. 

    Finance Festival - research sessions

    Make sure you register for the upcoming Finance Festival, to join sessions delivered by UWE on 'Building relationships between academics and post award teams' on Mon 16 November (morning) and UKRI's Rod Viggers on the latest funding assurance updates on Tuesday 17 November (afternoon). 

    Research Finance Teams Channel


    If you are not a member of the BUFDG Teams for research finance and would like to be, you can request access by completing the form on the knowledge hub, look for the link under 'what members are talking about'. We've added new channels dedicated to AI and facilities. , fill in form.

    Future meetings

    As we enter the third year of the Research Finance Forum's existence (yes, time really has gone that quick) we'll be taking stock to ensure the format and regularity is working well. It will also soon be time to replace our wonderful co-chairs Helen and James, who have been instrumental in building this community but will be excited to pass the baton for its next chapter. If you have any suggestions for the Forum (including collaboration with ARMA for those who are members) or an interest in supporting it going forward please contact Joni. We'll ensure regular attendees are kept up to date on future plans. 




    Read more



    This site uses cookies and other tracking technologies to assist with navigation and your ability to provide feedback, analyse your use of the site and services and assist with our member communication efforts. Privacy Policy. Accept cookies Cookie Settings