AI Agents for University Marketing in Australia: Enquiry to CRM

An AI agent covers the gap between the first enquiry and a usable CRM record

In Australian higher education, an AI agent is software that answers prospects on your website and other channels, collects what your recruitment team needs, and writes it into the CRM. It does not run your campaigns or decide who receives an offer. Its job is the gap where enquiries go cold: after the visit, before an adviser has seen anything.

Marketing pays for the traffic and admissions inherits the record. In between, a Year 12 student asks about a course late in the evening and gets a contact form. An agent answers first, then builds the record from the conversation.

This article follows the path from enquiry to CRM for Australian universities and TAFE-linked providers, shows what an agent can safely do at each step, and lists what to verify before it writes into your systems. For the wider picture, read our guide to AI agents for student recruitment.

The enquiry-to-CRM path in five steps

An agent can help at each of five steps without touching the offer decision.

Step three is usually where a weak setup shows. A chatbot that stores an email address gives the adviser a name and nothing to say. An agent that also records the course of interest, entry pathway and questions asked gives the adviser a way in.

Steps 1 and 2: answer first, then qualify

The agent should answer before it asks for anything, because a form-first experience loses the visitor. Australian prospects arrive with specific questions: what ATAR or selection rank does this course need, is there an alternative entry pathway, can I get credit for a diploma.

Answers must come from your own published content. Entry requirements and adjustment factors change, and an agent that improvises one creates a complaint. Link each factual answer to its source page, and let the agent say "I don't know, let me connect you with an adviser" rather than guess.

Keep qualification conversational. Three questions are usually enough:

• Are you finishing Year 12, coming from VET or TAFE, transferring, or applying from overseas?
• Which course or study area interests you?
• Which semester do you want to start?

Our guide to AI lead qualification for student recruitment covers scoring and routing. One caution applies from the start: qualification is a sorting step for follow-up. It must never become a hidden decision about who deserves an offer.

Step 3: write the record your CRM can use

A good agent record is small, consistent and traceable. The agent should write to the fields your team already uses.

Useful fields are course of interest, entry pathway, target semester, source page, consent status and a summary or link to the conversation. Free-text notes go in a labelled field so advisers can tell what the student said from what the agent inferred.

Three rules keep the CRM clean:

1. Limit the writable fields. List them in the contract and test that nothing else changes.
2. Match before you create. Check for an existing record by email, or you build the duplicates you later pay to remove. Our guide to cleaning a student lead database explains the clean-up.
3. Keep a log. Each write needs a timestamp and the conversation behind it so a bad record can be traced and reversed.

The CRM remains the source of truth. If you are still choosing a platform, our CRM comparison for higher education covers the main options.

Step 4: the handoff decides whether it paid off

A handoff works when the adviser opens the record and already knows what the student asked. It fails when they start over, which cancels the time the agent saved.

Test three things before launch. Does the whole conversation travel with the record? Does the student see when a person will reply? Does a request the agent cannot handle, such as a disability adjustment or a complaint, route to a named team and not a general inbox?

Open day bookings are the most dependable first action to automate. The agent offers a slot, logs it and sends the confirmation. Risk is low and the result is easy to check. Follow-up messages are similar: welcome when sent to someone who asked, a problem when sent to everyone.

Privacy: the Privacy Act, the APPs and the OAIC

The Privacy Act 1988 and the Australian Privacy Principles govern how most universities collect, use and disclose personal information. The Office of the Australian Information Commissioner publishes guidance, including on how the principles apply to AI products, and it is the first place to check before an agent goes live.

Five practices follow from the principles:

• Tell the prospect at the start that they are talking to an AI, and say what you do with their details.
• Collect only what the enquiry needs, and say why.
• Observe the Spam Act and Do Not Call Register rules before any follow-up message.
• Set a retention period for conversations and records, and apply it.
• Check where the provider hosts and processes data, including any overseas disclosure, and whether it trains models on your prospects' messages.

International recruitment adds a layer. Education agents and offshore prospects bring the ESOS framework and overseas privacy laws into play. If you recruit in the EU, Annex III of the EU AI Act lists systems that decide access to education as high risk. An agent that answers questions and books events is different from one that ranks applicants.

TEQSA, tertiary admission centres and honest claims

The agent informs prospects about applying and does not replace the application. State and territory admission centres such as UAC and VTAC handle many undergraduate applications, and your agent should point students there for the application itself.

The Tertiary Education Quality and Standards Agency regulates providers and expects accurate information for students. Treat every agent answer about entry, fees, accreditation or outcomes as a published statement made in your name, and have it draw from pages you have already approved. Consumer law also bars misleading claims, so an agent that overstates availability or graduate outcomes is a legal risk, not only a quality one.

How to measure it without inventing numbers

Measure the agent on the steps it controls, using your own data. Conversations that end in a complete record, adviser response time against your own target, and open day bookings are all countable today.

Enrolment is a long chain with many causes. Our guide to marketing attribution in higher education explains why crediting the agent with every enrolment misleads. Run it through a full intake cycle, compare against the same period a year earlier, and say what else changed. For channel context, see the digital marketing guide for higher education.

Five questions to ask any provider

Ask these before you sign, including of us.

1. Which CRM fields can it write, and can we restrict them?
2. Can we see a log of every write and undo one?
3. Where is data processed, and is it used to train a model?
4. How is the handoff tested on real conversations?
5. What does the agent say when it does not know?

Be sceptical of any case study with a headline percentage and no method. No public benchmark defines a normal result for a recruitment agent.

Where Skolbot fits

Skolbot deploys AI agents connected to the institution's CRM. The website agent answers from the institution's own content and passes the conversation and its qualification into the CRM. The CRM remains the source of truth, and the admissions team stays in control of the cases that need it.