CASE AI Coffee Chat: Practical Uses of AI, from Due Diligence to Automated Reporting-Recording

Video Transcription

Video Summary

The session opened with housekeeping about Q&A, chat questions, and timing, then featured two case studies on using AI in higher education advancement.<br /><br />Aidan O’Kane from Ulster University described how his team used agentic AI to automate donor due diligence. The manual process took about 3.5 hours per person and around 84 hours per year at low volume, with staff gathering public information from sources like Companies House, sanction lists, Wikipedia, and news articles. Using Doozer.ai, a staff member sends a simple email request to a digital co-worker, Atlas, which compiles a report in 5–15 minutes. The system keeps a human in the loop for final judgement, avoids internal system integration, and uses only public data, reducing privacy concerns. Aidan reported a 97–98% reduction in effort and said the proof of concept has encouraged wider use across the university.<br /><br />Charlotte from RCSI presented a more DIY approach, using AI as a coding partner to automate the CASE Alumni Engagement Metrics (AEM) report. With no formal data training but some Python knowledge, she used ChatGPT to generate and refine code that checks whether contacts appear in volunteering, philanthropy, communications, and experiential datasets. She emphasized good prompting, testing, and debugging, plus avoiding real customer data in generic AI tools. The project took about 25 hours over three weeks, and future annual updates should take around 30 minutes. Charlotte said the process improved her technical skills and could be reused for other reporting and prospecting tasks.<br /><br />The session closed with CASE resource links, upcoming webinars, conference announcements, and encouragement for attendees to contact the speakers.

Keywords

higher education

artificial intelligence

donor due diligence

agentic AI

automation

fundraising analytics

Alumni Engagement Metrics

data reporting

ChatGPT coding

public data