An education assistant that ignores the course and answers from the open internet will teach the wrong thing with confidence. The useful version is grounded in the materials that course actually uses, lives inside the LMS people already open, and refuses questions it cannot support from those materials.
What makes AI work different here
- The course is the source. Answers should cite the lesson, the syllabus, or the reading. A general answer that contradicts the course is a defect.
- The LMS is old on purpose. Schools and training teams will not rewrite a five-year-old platform for one feature. Integration is the job.
- Student data is not training fodder. Grades, messages, and attempts stay in the scope you set. We do not treat the gradebook as free fine-tuning data.
- A tutor is not an examiner. Help and hints are a different product from grading. Grading needs a rubric, a record, and a person where the result matters.
Where we typically start
We start from the LMS, the content the tutor is allowed to use, and what the assistant must not do, including grading and inventing policy. The published example added a tutor to a legacy LMS without replacing it.
Services we typically provide
Case studies from this industry
Frequently asked questions
Will this grade students?
Not as an automatic final mark. Help grounded in the course is the scope. Grading needs a rubric and, where the result matters, a person.
Does it need a new learning platform?
No. The published work added the tutor to a five-year-old LMS in five weeks without replacing the core.
Can it answer outside the course?
It should say when the course materials do not cover the question, rather than fill the gap from general knowledge.