AI in admissions: why strong processes come first

Your admissions team is probably already using AI. The question is whether your processes are ready for it.

At EnrolyCon26, Kylie Evans, Head of Admissions Operations at Aberystwyth University and Meg Medlicott from Enroly explored how universities can use AI in admissions without losing control, consistency or compliance.

The central message was clear: do not use AI to compensate for a fragmented process. Build strong operational foundations first.

That principle is also behind Enroly Apply. By helping universities capture more complete, structured applications and create clearer workflows, Apply provides stronger foundations for admissions decisions and the responsible adoption of AI.

Start with the process, not the tool

The best AI use cases often begin with a specific operational frustration.

For Meg, the challenge was finding information quickly across multiple folders, case studies and internal documents. For Kylie, it was tuition fee assessment, a complex process involving regulations, residency checks and significant room for inconsistency.

Neither started by asking, “How can we use AI?”

They began by identifying a repetitive, time-consuming process that could be improved.

Useful starting points for admissions teams could include:

• Summarising policies and guidance.
• Highlighting missing application information.
• Supporting consistency checks.
• Finding relevant internal information.
• Drafting process summaries.

Starting with a defined problem makes it easier to test whether AI is genuinely adding value.

Keep people in control

AI should support decisions, not make them independently.

Kylie’s custom GPT could suggest a likely fee-status outcome, identify missing information and explain the reasoning behind its recommendation. However, the final decision always remained with the admissions team.

AI can summarise, recommend and flag possible issues. People must still interpret nuance, review exceptions and remain accountable for the outcome.

For admissions and compliance teams, human oversight is not an optional extra. It should be designed into the process from the beginning.

Build clear guardrails

AI outputs should never be treated as automatically reliable.

Meg shared a simple traffic-light confidence system:

• Green: high confidence.
• Amber: incomplete information or partial confidence.
• Red: unclear or requiring human review.

This makes uncertainty visible rather than allowing every answer to appear equally authoritative.

Kylie also restricted her tool to approved regulations and controlled information sources. Applicant data used during testing was anonymised.

Before scaling any AI use case, universities need clarity around:

• Approved sources.
• Appropriate data use.
• Mandatory human review.
• Accountability for decisions.

Better prompts need clearer processes

Vague instructions tend to produce vague results.

Meg demonstrated how outputs improved when prompts included a defined role, context, constraints, approved sources and a clear output format.

For example, rather than asking AI to “summarise this policy”, a stronger instruction would ask it to act as an admissions compliance reviewer, identify operational risks and highlight areas requiring human review.

This is not only about learning to write better prompts. To instruct an AI tool effectively, an institution first needs to understand how the process should work, which rules apply and what a good outcome looks like.

Start small

Universities do not need to automate an entire admissions journey to begin exploring AI.

The strongest early use cases are often small, controlled and easy to review, such as internal knowledge tools, policy summarisation or consistency checking.

The objective should not be AI transformation overnight. It should be safely reducing one defined source of operational friction, testing the outcome and learning before expanding further.

Strong admissions processes come first

AI is already entering admissions teams, whether institutions have formal guidance in place or not. The challenge is therefore not simply whether to adopt it, but how to use it responsibly.

The institutions likely to gain the greatest value will be those applying AI to processes that are already structured, consistent and understood.

That means creating strong foundations:

• Complete and structured information.
• Clear requirements.
• Consistent review.
• Shared visibility.
• Defined next steps.
• Appropriate governance.

This is the thinking behind Enroly Apply.

Apply helps universities capture more complete application information and evidence, coordinate review and create a clearer journey for applicants, agents and admissions teams.

Because AI is not the starting point. Strong admissions processes are.

🔗 Explore Enroly Apply here.

‍