Operations & Workflow Management
AI Applications in Research Administration Operations
Free open AI tools could be used with non-confidential information to strengthen communications and other plans, create onboarding and training resources, and complete custom workflow tasks.
Research administrators (RAs) face pressure to increase efficiencies to help address federal funding and compliance changes, broader issues facing higher education, and shrinking resources. The authors explore applying free open AI solutions to some common operations tasks. RAs could use free tools with care not to breach confidentiality (Schultz, 2025b).
Creating a Customized Profile
ChatGPT can be customized for response preference using the user's profile so that outputs such as award summaries are closely aligned with the user’s communication style, professional and courteous. By providing the user's professional information and desired communication style and format, the user can create more personalized information relevant to research administration and the user’s organization. Additionally, a user can direct the AI not to invent information and to instead explicitly state when it doesn't know the answer (Harmon, Matronick, Robison, & Bachman, 2025) and further instructions, for example, about terminology such as “Principal investigator and project director are synonymous.” These types of instructions would help with repeated queries, such as summarizing award notices. A query with leading questions about what to include in the summary could be used, and instructions in the profile would clarify ambiguous terms. The award summaries could be used for creating internal institutional announcements of the awards. All queries would automatically be responsive to the RAs work and organization if these details are included in the profile.
Strengthening a Communications Plan
ChatGPT can organize an outline of a communications plan into a formatted plan and add new ideas for sharing information. The communications plan could identify the audience for sharing grant resources and information, methods to reach the audience, and a content strategy for what to share. From a bulleted list and a prompt to strengthen the writing, ChatGPT is able to enhance the plan. The communication plan can be converted into a sharable slide deck for meetings. However, ChatGPT’s output is only as good as the prompt that it is provided with. The more specific the prompt the more detailed the answer will be. A great prompt includes role, context, task, and format (Harmon, et al, 2025). The role and context could be provided in a user's profile, or they could be included in a prompt. The task is what action is requested, such as summarize or strengthen the writing. The format is what type of output is needed, such as text, numerical analysis, slides, or graphic design. AI can even help with strengthening prompts.
Onboarding and Training
We tested OpenAI (2025) for onboarding new staff and training. ChatGPT could create materials providing overview and background of the field. A prompt to provide an overview of research administration for a new clerical support position resulted in an overview. ChatGPT offered to create a visual diagram of the proposal workflow and one page reference guides for a desk handbook. Also, ChatGPT had great ideas in response to a prompt to create a board game about the grant lifecycle.
Creating a Custom Workflow
A custom task or workflow can be created in ChatGPT, using the “GPTs” button on the left panel, then choosing “Create” on the top right. A workflow could be created to identify problematic contract terms and conditions (Schultz, 2025a). It could also be used to create award summary sheets for principal investigators (Schultz, 2025a). An award summary sheet is more of a checklist of highlighted terms and conditions that could be used in a grant kick-off meeting, rather than the award summary that could be used to publicize the award.
The University of Idaho received an NSF GRANTED award to develop AI tools that could be shared with research administrators (University of Idaho, 2024). As a part of this effort, they have published “Practical Guide to Using AI in Research Administration: the TaMPER Framework and the AI Workflow Design Worksheet (https://ai4ra.uidaho.edu)” online and plans to release more resources as they are developed. These resources guide creation of custom workflows.
ChatGPT 5 now has an advanced deeper reasoning (GPT Thinking) model. To ensure the AI is responsive to more complex requests, (and to ensure the thinking model is used), include “Think hard about this” in your prompts.
Challenges
The challenge is how one checks the accuracy of AI outputs if the user is a beginner in the field. The user must check the source and use multiple sources to cross-check the validity of the output when available. As AI tools develop, original documents can be cross-referenced with links in the AI output results, so users can check summaries with original documents (Harmon, et al, 2025). The University of Idaho is using the TaMPER Framework to develop task workflow assignments that provide linked output alongside the original source for verification.
Free open AI tools could be used with non-confidential information to strengthen communications and other plans, create onboarding and training resources, and complete custom workflow tasks. Watch the University of Idaho AI4RA project as tools develop. The Federal Demonstration Partnership may also provide tools and resources in the future (Schultz, 2025a).
References
Harmon, D., Matronick, S., Robison, B., & Bachman, B. (2025, August 10). Harnessing AI to tame research administration complexity [Conference presentation]. NCURA 67th Annual Meeting.
OpenAI. (2025). ChatGPT (GPT-5) [Large language model]. https://chat.openai.com
- Schultz, L. A.
(2025, July). Award receipt/negotiation/management – Using AI [Video]. NCURA YouTube Tuesdays. https://www.youtube.com/watch?v=hGQuysNui7M&t=115s
- (2025, August). AI: Research administrators meeting the moment. NCURA Magazine, 57(4), 21.
University of Idaho. (2024). Artificial intelligence for research administration. https://ai4ra.uidaho.edu
Disclaimer: The views presented in this article do not represent those of any organization. The information provided in this article is for general informational purposes only and does not constitute professional advice. The authors make no representations or warranties of any kind, express or implied, about the completeness, accuracy, reliability, suitability, or availability with respect to the article or the information, products, services, or related graphics contained in the article for any purpose. Any reliance you place on such information is therefore strictly at your own risk.