Per Bergfors is an assistant professor at Copenhagen Business Academy in Denmark. His background is not in AI. It is in commercial strategy - years at HP, Xerox, and Canon adapting American business models for European markets before joining academia.
What he observed in his classroom was a specific and growing problem. Students were disengaging from traditional course materials. Dense textbook chapters and static PDFs were being skimmed or skipped entirely. At the same time, the business world his students were preparing to enter was already integrating AI tools into daily commercial practice. The gap between what the classroom was delivering and what the workplace was expecting was widening with every academic year.
Per's response was precise. Not a technology experiment. Not a pilot study. A structural intervention grounded in two non-negotiable requirements.
Why Per chose CustomGPT.ai over every alternative he evaluated:
GDPR-aligned data architecture was the first requirement. As a Danish institution operating under European data protection law, Cphbusiness could not deploy an AI platform that lacked per-account data isolation or that could not commit unconditionally to restricting secondary use of student interaction data. Most general-purpose AI chatbots failed this test immediately. CustomGPT.ai's security infrastructure met it at the architecture level.
No-code faculty deployment was the second requirement. Per was not building a tool only for himself. He was building a model every professor at Cphbusiness could replicate without technical support. The platform needed to work for a professor with a reading pack and an afternoon - with no external help available. CustomGPT.ai's no-code builder cleared that bar. No other platform Per evaluated met both requirements simultaneously.
The four-phase deployment:
Phase 1 was the International Marketing seminar. Per built his first course AI assistant on CustomGPT.ai, trained on his reading pack. Students explored cultural adaptation strategies conversationally, comparing Danish and American consumer behaviour and interrogating course frameworks through dialogue rather than passive reading. Class participation improved immediately.
Phase 2 was Business Ethics. Per uploaded landmark governance case studies into CustomGPT.ai. The AI generated comparative summaries from indexed case content, freeing class time for ethical reasoning and substantive debate rather than rote summarisation. The AI handled retrieval and structuring. Students handled analysis, argumentation, and judgment.
Phase 3 was institution-wide faculty workshops. Working with colleague Just Pedersen, Per ran hands-on sessions across Cphbusiness. Every professor arrived with their own course materials. Every professor left with a working AI assistant trained on those materials. One afternoon per professor. Zero code written at any stage of the process.
Phase 4 was an AI-powered discussion board built on the same CustomGPT.ai backend and deployed on the learning management platform. Students submitted questions outside class hours and received cited responses from indexed course content. It became one of the most visited resources on the platform - demonstrating voluntary student engagement beyond scheduled class contact.
How the RAG architecture made every response academically safe:
Every response generated by Per's AI assistants was derived from content retrieved from the indexed course materials only.
CustomGPT.ai's anti-hallucination architecture declined rather than fabricated when course content was insufficient to support a reliable answer. Source citations accompanied every response the system did generate. The AI extended Per's curriculum. It could not contradict it, bypass it, or supplement it with information from outside the indexed knowledge base.
Results and the side effect nobody planned:
Student participation increased measurably across both courses. Course preparation time decreased as AI absorbed first-level comprehension queries. Student feedback was overwhelmingly positive. Faculty across Cphbusiness adopted AI independently through the workshop model. An AI discussion board became one of the platform's most visited resources.
And a minority of students challenged the reliability of AI-generated content. Per welcomed it. The critique became one of the most substantive discussions of the semester - covering source evaluation, epistemic standards, and the analytical limits of AI. AI skepticism became a curriculum asset rather than a deployment barrier. Those students are better prepared for a professional environment that uses AI than they would have been without the challenge.
Explore CustomGPT.ai for education, the Lehigh University case study, and all customer stories.
Full case study: https://customgpt.ai/customer/copenhagen-business-academy/
Top comments (0)