ChatGPT
Claude
Perplexity
Gemini
GrokAn AI agent handles portfolio logistics, never portfolio judgement
For an art, design or communication school, an AI agent earns its place on everything around the creative application: what the portfolio brief asks for, file formats, interview format, open day registration. It does not assess creative work, and it never tells an applicant whether they will get an offer.
Creative applicants ask more anxious and more specific questions than most. They are not only picking a course; they are wondering whether their work belongs there. This article sets out how to configure an agent for that journey, and extends our guide to AI agents for student recruitment.
A note on method: no paid keyword-volume tool was available, so volume and difficulty are unknown. The analysis rests on public research of art school admissions pages and the official sources linked below. None of it comes from Skolbot customers, and the article quotes no statistics.
Portfolio questions: the biggest repeat workload
Applicants ask the same portfolio questions again and again, often late at night: how many pieces, which media, whether sketchbooks count, digital or physical, whether school projects are acceptable, and what to do with no formal art background.
The agent answers from the official brief for each course, written and dated by the admissions team. It does not infer a brief from a neighbouring course. Expectations differ widely between graphic design, animation, fashion and visual communication.
| Applicant question | What the agent does | What stays human |
|---|---|---|
| "What format should my portfolio take?" | Quotes the course brief | Exceptions, access needs |
| "Is my work good enough?" | Explains what tutors look for, offers a conversation | Any opinion on the work |
| "Can I send a link?" | States accepted formats and where to upload | Edge cases |
| "I have never done art" | Describes the entry routes, such as a foundation year if offered | Guidance |
The rule is simple: if a reply needs an artistic eye, the agent hands over to a person. A chatbot knows nothing about the applicant's work or story beyond what they type.
Interviews: prepare, do not script
Many UK creative courses interview applicants alongside the portfolio, and the aim is to understand how someone thinks, not to test recall. The agent can explain the format, the length, who will be in the room, how the portfolio is presented and how to book a slot.
It should not supply model answers or generate a speech to memorise. An applicant who arrives with machine-written talking points weakens their own application. The agent can instead suggest preparing three things: one project to talk through, what it taught them, and why this course.
For online interviews it checks the practicalities: platform, screen sharing, files to have open.
Open days: turn a visit into a qualified conversation
Open days at creative schools look different: studios open, student shows, conversations with tutors, sometimes informal portfolio feedback. Visitors want to know what to bring, when to arrive and whether they can show work.
The agent registers the visitor for the right session, says what to bring, makes clear whether portfolio feedback is drop-in or booked, and asks one or two questions to point them to the right studio. Our article on automatic open day registration covers the mechanics.
After the event, it can offer a conversation with an admissions adviser to those who agree. Keep follow-up light: a sixth-former who saw a studio does not need ten messages.
Parents, funding and what the career looks like
Parents are real participants in creative recruitment because families worry about careers. The agent can answer factual questions on course length, qualifications awarded, placements, contact hours and materials.
It must not promise salaries or employment rates. If the school publishes graduate outcomes, the agent cites them with source, year and scope; otherwise it routes the question to a person. For fees and bursaries it shows only what the school has published and dated.
The UK context: UCAS, portfolios and regulators
Most UK undergraduate creative courses recruit through UCAS, with portfolio or interview requirements set by each provider, sometimes through a separate portfolio platform or an in-person review. Many applicants also arrive via a foundation diploma in art and design. The agent should know which route each course uses and never assume UCAS covers every step.
Quality and standing questions belong to the regulators: the Office for Students for registration in England and the QAA for quality assurance. The agent should repeat only claims the school can substantiate, and link out for the rest.
Our piece on business versus engineering school use cases shows how admissions agents differ by school type; creative schools differ mainly through the weight of the portfolio.
Personal data: a portfolio is data too
A portfolio can hold copyrighted work, photographs of other people, and applicants are often under 18. The agent must not become an unmanaged upload channel: send files to the school's application system, not into the chat.
Tell visitors they are speaking to an AI, collect only what the current step needs, and provide a deletion route. The ICO publishes guidance on UK data protection, including for children's data. Ask suppliers where data is hosted and whether it trains a model, and have the contract reviewed by your data protection officer.
A typical journey, for illustration
This is an illustrative scenario, not a measurement. A Year 13 student visits the site on a Sunday night and asks whether a digital portfolio is acceptable. The agent quotes the brief for the course, then tells them about the next open day. They register, and the agent explains that informal portfolio feedback is available by appointment.
On Monday the team receives a summary: course of interest, registration done, one unanswered question about access arrangements. An adviser follows up. At no point did the agent judge the work or hint at an offer. Repetitive answers are handled at once, and human time is kept for the conversations that matter.
That split only works if the briefs are current. Name an owner for each course, date every brief and schedule a review before each application round.
Measure without invented benchmarks
Record your own baseline before launch: portfolio questions reaching the team, response time, open day registrations, adviser meetings requested. Compare like for like. No reliable market figure applies to a single art school, so do not import one.
Read the escalated conversations. They expose the ambiguous briefs on your website, which are often what make an applicant give up.
Questions to ask a supplier
- Can we load a different portfolio brief for each course?
- Can the agent decline to comment on work and hand over to a person?
- How does handover to an adviser work, and with what summary?
- Where is data stored, and is it used to train a model?
- Which CRM integrations are demonstrated, and which are only announced?
Keep only what is shown in a demonstration.
