How Elon stood up TUEL-powered tutoring for Spring 2026

In December 2025, Elon University stood up TUEL-powered tutoring for the Spring 2026 semester. The first live deployment was not a campus-wide program. It was a focused implementation in Dr. Mustafa Akben's business courses, shown to students as Elon AI.

This post is a launch note. It records what was put in place before the semester began. It does not report semester outcomes. Those belong to later pages, written after Spring 2026 had actually run.

Why Elon stood this up

Mustafa needed a tutor that would answer from his own course materials rather than the open web. Students already had generic chatbots. The problem with those tools in a credit-bearing course is simple: a fluent answer is not the same as a course-aligned answer, and faculty cannot check a source that was never shown.

The December work was therefore about standing up a course-grounded tutor in time for Spring 2026, with faculty controls in place and citations treated as a design requirement. The scope was Mustafa's business courses. That is the honest frame for this dateline.

How the implementation was stood up

The first job was material upload. Mustafa uploaded the materials students would actually be assigned — textbooks, lecture slides, and supplementary readings — so the tutor would index a course-specific knowledge base instead of a generic corpus. The point of that step is narrow and practical: when a student asks a question in these courses, the assistant should work from what the instructor assigned.

Faculty controls sat next to that corpus. The instructor could configure how the tutor behaved for the course, set response boundaries, and keep the assistant aligned with the syllabus. Students would meet the same system under the Elon AI name. They were not being handed a generic vendor chatbot with a campus sticker on it.

Citation-as-design was the third piece. Every response was expected to point back to the uploaded materials so a student or instructor could check the source. That is the difference between a plausible answer and a teachable one. It is also the control that lets faculty review the assistant without treating it as a black box.

None of this required a multi-department rollout. The December launch was a working tutor in a defined set of business courses, ready for the term that was about to start.

What this note does not claim

This post is dated December 20, 2025. On that date, Spring 2026 had not been taught yet. There are no adoption rates, exam averages, session counts, token totals, error rates, or satisfaction scores in this note, because those figures did not exist yet for this deployment.

Later reporting from the same Elon deployment lives on the Spring 2026 proof pages. This URL stays a launch note so the December dateline does not pretend to be a completed-semester case study.

For the Spring 2026 semester outcomes from this same Elon deployment, read the Spring 2026 Elon University case study. A longer write-up of that semester is also on the blog: Elon University's results after one semester with AI-powered course tutoring.

What the launch left in place

Elon AI is the student-facing name for this deployment. TUEL — Trusted Unified Education & Learning — is the platform underneath it. The December launch put a course-grounded tutor, with material upload, faculty controls, and visible citations, into Mustafa's business courses in time for Spring 2026.

What students and faculty did with that tutor over the semester is a separate story. It is told after the term, on the pages linked above, not on this dateline.