An AI chatbot on a university website has to deal with the same thirty or so questions, over and over, in every cycle. This list sets them out in six groups and says, for each one, whether the agent should answer from the institution's own published content, answer and then collect some context, or pass the conversation to a person. The principle behind all of it is simple: answer from what the university has published, collect context along the way, and hand off whatever needs judgement.
If you are still deciding whether to deploy one at all, start with our guide to AI chatbots for student recruitment and the piece on what AI agents do across the recruitment journey. This article is the practical checklist that follows.
A. Programmes and entry requirements
Most first conversations start here. Every answer should come from the course page, not from the model's general knowledge, because entry requirements differ course by course.
1. Which course or level is right for me? The agent asks about interests and current qualifications, then points to matching course pages. It does not rank courses by quality, and it passes the conversation on if the applicant wants a personal recommendation.
2. Do I meet the entry requirements? The agent quotes the published requirements for the course: A-levels, BTECs, Scottish Highers, an Access to HE Diploma, or equivalent. It does not predict an offer. Contextual or borderline cases go to admissions.
3. How do I apply, and what are the steps? For most undergraduate courses the answer is UCAS, so the agent explains the route and links to the UCAS website. Postgraduate and some part-time courses use a direct application, which the agent should describe from the course page.
4. What are the deadlines and start dates? The agent reads the dates from the course page and says plainly when it cannot find one. If a course is still open late in the cycle, it can explain how clearing works at that university, as long as the page says so.
5. Is there an interview, audition, portfolio or test? The agent describes the selection stage published for that course, such as an interview, a portfolio review or an admissions test. Requests to prepare someone individually, or to reschedule, go to a person.
B. Fees and funding
Money questions arrive early and often. The agent's job is to point at the published fee schedule and the official funding bodies, not to quote figures from memory.
6. How much is tuition? The agent gives the fee shown on the course page for the right fee status, home or overseas, and states which academic year it applies to. If the status is unclear, it asks one clarifying question first.
7. Can I get a scholarship or bursary? It lists the awards the university publishes and their published criteria, then links to the application page. Eligibility judgements on a specific person belong to the student funding team.
8. How do I pay, and when? The agent explains the published payment options and instalment dates. A request for a different arrangement is a hardship or exception case and is handed off (see question 26).
9. Can I work while I study, or take a degree apprenticeship? The agent explains what the university publishes about part-time work, placements and degree apprenticeships. For tuition support from government, it points to Student Finance on GOV.UK rather than summarising rules itself.
10. What else will I pay for? Accommodation, field trips, equipment, printing and professional body fees vary by course. The agent cites the course's own cost notes and says so when the page does not list them.
C. Recognition and outcomes
Applicants and parents want proof that the qualification counts. The agent should send them to the university's own dated figures, never to a number it invents.
11. Is the university and the degree properly recognised? The agent states whether the provider appears on the Office for Students register and links to the OfS website. It can also mention QAA review reports where the university publishes them. Questions about a professional body's recognition of a course go to the course team unless the page answers them.
12. What happens to graduates? The agent cites the outcomes data the university publishes, with the survey name and year, such as the national Graduate Outcomes survey. It does not turn that into a promise.
13. Are there placements or internships? It describes the placement year, work experience or industry projects listed on the course page, and says whether they are optional or built in.
14. Can I study abroad or take an exchange? The agent reads the partner and exchange information on the international office page and says which courses are eligible.
15. Where does the university sit in the league tables? It repeats only what the university publishes, with the table name and year, for example the Guardian or the Complete University Guide. It does not compare itself to named rivals.
D. Campus and practical life
These questions are easy to answer well and make the biggest difference to a prospective student's first impression.
16. Where will I live? The agent explains university halls, application timing and the private rental process from the accommodation pages. It hands off special requests, such as accessible rooms, to the accommodation office.
17. Where is the campus, and how do I get there? It gives the address, transport links and directions published by the university, and can explain how a city-centre or out-of-town site works day to day.
18. Can I visit? How do I book an open day? The agent shows upcoming open days and campus tours, takes the booking and captures the course of interest. This is one of the clearest places where the conversation can pass into the CRM with context.
19. What is the timetable and teaching format? It describes whether delivery is on-campus, blended or online, and typical weekly patterns where the course page says so. Individual timetable queries from enrolled students belong to a different channel.
20. What is student life like, and what support exists? The agent summarises societies, the students' union, careers, academic and wellbeing services from the published pages. It does not describe a person's specific entitlement.
E. International applicants and paperwork
International questions need accurate official sources, because a wrong answer can have immigration consequences.
21. Do I need a visa? The agent explains that most international students need permission under the Student visa route and links to GOV.UK's student visa guidance. It does not assess an individual's eligibility.
22. What English language level do I need? It quotes the course's published language requirement and the list of accepted tests. Questions about a test result that falls short go to admissions or the international team.
23. Will my overseas qualification be accepted? The agent explains how the university checks qualifications and which equivalency information it publishes by country. The decision on a specific certificate is made by admissions.
24. Which documents do I need? It lists the documents the university publishes, such as transcripts, references, passport details and a personal statement, and explains where they are uploaded.
25. What is the status of my application? The agent cannot answer this without authenticated access to the admissions system. Without it, it explains the usual stages and passes the request to a person, along with the reference number the applicant supplied.
F. Questions that must go to a human
Some questions should never be settled by an agent. The right behaviour is a quick, kind handoff with the context already attached, so the applicant does not have to repeat everything.
26. Financial hardship or a fee arrangement. The trigger is any request to change a fee, delay payment or discuss difficulty paying. The agent passes the course, fee status and the applicant's stated situation, and tells them who will reply and when.
27. Special circumstances and adjustments. This covers disability, health conditions, a gap year, a failed year, or interrupted study. The agent shares the support-service contact and passes the course and any details the person chose to share. It never asks for medical information.
28. Complaints, appeals and rejection decisions. The agent acknowledges the issue, provides the published complaints or appeals route and hands over the conversation. It does not argue the decision or predict the outcome.
29. Wellbeing, safeguarding or an emergency. If someone mentions harm to themselves or others, the agent gives the published urgent-help contacts straight away and alerts a person. This trigger overrides every other flow.
30. Data rights requests and sensitive personal data. A subject access request, an erasure request or a complaint about data handling goes to the data protection officer. The agent does not collect extra personal data to process it and passes only the request itself.
How the thirty questions are routed
| Theme | Answer autonomously | Answer, then collect context | Hand off |
|---|---|---|---|
| A. Programmes | Requirements, steps, dates | Course interest, qualifications | Offer predictions, tailored advice |
| B. Fees and funding | Published fees, awards, funding links | Fee status, course | Eligibility rulings, payment changes |
| C. Recognition and outcomes | Register status, published outcomes | Career interest | Professional body queries |
| D. Campus and life | Location, formats, support services | Open day booking, course | Accessible rooms, special requests |
| E. International | Visa overview, language levels | Country, intake | Qualification checks, application status |
| F. Human only | Nothing beyond signposting | Minimal context to pass on | Everything |
How to test the agent before launch
Run the thirty questions as a written test script, using the exact wording applicants use rather than the tidy versions above. For each answer, open the page it should have come from and check that the facts match. If the agent says something the page does not, treat it as a failure, however plausible it sounds.
Add questions the agent should not be able to answer, such as a request for a guaranteed offer, and confirm that it says it does not know and offers a person. Then test the six handoffs: check that each one fires, that the right team receives it and that the context arrives intact.
After launch, review the questions the agent could not answer every month. Each gap is either a missing page, which the university should write, or a question that rightly belongs with a human. Fix the content first, because the agent is only as good as what the university has published. Our notes on the questions prospects ask before enrolling are a good source for extra test cases.
Disclosure and data protection for UK universities
Tell people they are talking to an AI at the start of the chat, and make it easy to reach a person. In the UK, personal data in chat is covered by UK GDPR and the Data Protection Act 2018, and the regulator is the Information Commissioner's Office. The ICO publishes a hub of guidance on AI and data protection that is a sensible starting point for a privacy notice and impact assessment.
If you also serve applicants in the EU, Article 50 of the EU AI Act requires that people are informed when they interact with an AI system; the text is on artificialintelligenceact.eu. Collect only what the question needs, state the purpose, and keep special category data out of the chat. Your data protection officer should sign off the flow before launch.
Where Skolbot fits
Skolbot's web agent answers using the school's own content, understands programmes, intakes and campuses, and passes the conversation and its qualification to the CRM so the admissions team takes over with context. It sits alongside advisors, not in place of them, and the comparison in AI chatbot versus human agent explains how to divide the work.



