Give the chatbot a bounded job
A high school capstone can involve dozens of different internships, making individualized weekly feedback difficult to sustain. This design report describes a chatbot configured around course rubrics, requirements and learning theories to help seniors set goals and reflect on obstacles. Teachers compared four configurations and simplified an opening prompt that had become overwhelming. The final design concentrated on weekly planning and reflection, reserving project approvals and policy decisions for school personnel. The model used instructions and approved documents, rather than fine-tuning on student data. The contribution is a concrete account of instructional design and its unresolved safeguards; the report does not demonstrate improved achievement, reduced workload or effective protection of minors.
Why it matters for leaders
Define one narrow support function before commissioning a school chatbot. Require a route to a named adult for approvals, uncertainty and sensitive disclosures. Review the documents it uses, test incorrect policy questions, and measure whether its prompts improve the next human advising conversation.
The evidence
An administrator, two teachers and three researchers developed the low-code system. Course alignment and teacher feedback guided iteration; learning outcomes remain future work.
Who was studied
Seniors at a university-affiliated public high school completing community internships.
What the evidence cannot tell us
No controlled outcome evaluation. Safety testing required repeated revisions, and the authors describe continuing uncertainty about data use and safeguards for younger learners.
Read the original sources
Summary and leadership analysis from the Week ending 2026-10-09 brief. The edition’s selection notes explain source access and evidence limitations.