Education & nonprofit · website AI assistant
AI chatbot for Education
Updated
The short version
Handle admissions, deadline, and course questions from your prospectus and policy pages — in the applicant's own language, with sources.
Education websites carry an enormous amount of correct information that almost nobody can find. Entry requirements, term dates, fee structures, visa guidance, module lists — all published, all buried three levels deep in a site built by committee.
The result is predictable: an admissions inbox full of questions the website already answers, spiking exactly when staff are least able to absorb it, and often arriving from applicants in a different time zone writing in their second language.
MySiteGPT trains on the prospectus, course pages, and policies and answers those questions directly, citing the page it used so an applicant or parent can check the official wording themselves. When the answer isn't published, it says so rather than improvising an entry requirement.
What schools, universities, and course providers get asked
Real questions, in visitors' own words. Every one of these is answerable from content you probably already publish.
- “What are the entry requirements for this course?”
- “When is the application deadline, and what happens if I miss it?”
- “How much are fees for international students?”
- “Do you accept this qualification from my country?”
- “What accommodation options are there in first year?”
- “Who do I contact about accessibility support?”
How MySiteGPT helps schools, universities, and course providers
Absorbs the admissions-season spike
The same fifty questions arrive thousands of times between now and the deadline. Answering them from your published pages frees staff for the cases that genuinely need judgement.
Answers in the applicant's language
International applicants ask in their own language and get a reply in it — 50+ supported — still grounded in your English course and policy pages.
Cites the official page
Entry requirements and fees are decisions people make life plans around. Every answer links the source page so nobody is acting on a paraphrase.
Refuses to guess about eligibility
When a qualification or a case isn't covered by your published criteria, the assistant returns your fallback and captures an email for the admissions team instead of implying an outcome.
What to train it on, in order
The answers are only as good as the index behind them. Start at the top — these are ordered by how much question volume they absorb per page trained.
- 1
Course and programme pages
Module lists, durations, and entry criteria — the highest-volume questions in the whole set.
- 2
Fees, funding, and scholarship pages
Train the published figures verbatim, and use Q&A pairs where the answer depends on status so the phrasing stays precise.
- 3
Term dates, deadlines, and the academic calendar
Small pages, huge query volume. Re-sync them when the calendar is updated.
- 4
Student handbooks and policy PDFs
Upload them directly — most of this content was never published as web pages and is invisible to search.
Not sure your pages are ready to be retrieved from? Run one through the RAG readiness checker first, or audit the whole site with the sitemap analyzer.
What to measure once it's live
We don't publish other people's numbers as if they were yours. Track these instead — they are all visible in your own dashboard within a fortnight.
- Admissions inbox volume during application season
- After-hours questions answered, by time zone
- Unanswered questions about eligibility — usually an unpublished criterion
- Enquiries captured as leads for the admissions team to follow up
AI chatbot for Education: frequently asked questions
Can it give an admissions decision or assess an application?
No, and it's designed not to. It explains published criteria and cites them. Anything requiring judgement about an individual case hits the fallback and turns into a captured enquiry for a human.
Is it safe to use with prospective students under 18?
It answers from your published content and doesn't require an account to use. Treat lead capture the way you treat any web form: ask for contact details only where your safeguarding and privacy policies allow, and link that policy in the widget copy.
Can different departments have their own assistant?
Yes — the plan includes three bots, so a common split is admissions, current students, and one faculty or department with distinct content.
How do we keep it accurate across an academic year?
Sources re-sync daily at a time you set, so calendar and fee updates flow through automatically. For a big annual change, trigger a re-sync immediately and spot-check the affected questions in the playground.
Try it on your own content first.
The chat bubble on our homepage is the product, trained on our own site — ask it anything and watch every answer cite its source. Then train one on yours: 14 days, full features, no card.