ChatGPT
Claude
Perplexity
Gemini
GrokThree tools, three jobs
An AI admissions agent converses with a prospective student and answers in real time. A simple chatbot returns prepared answers to expected questions. CRM automation runs rules and workflows on records already in your system. Compare them by the job each does, not by feature lists.
Canadian recruitment teams face an added wrinkle: applicants arrive through provincial systems, often in two languages, and from abroad. A student in Ontario asking about entry requirements at 10pm needs an answer now. An international applicant who attended a virtual session needs a timely follow-up. Different moments, different tools.
This guide gives a decision matrix, suggests ownership and sets out the federal and provincial rules that shape both approaches. It extends our pieces on AI agents from first contact to CRM and the CRM comparison for higher education.
Definitions
- FAQ chatbot. Follows a scripted tree or matches canned answers. Predictable, and brittle when a question is phrased differently.
- AI admissions agent. Understands free-text questions, answers from content the institution has published, asks qualifying questions, offers an appointment or information session and hands over to a staff member. See the glossary of admissions AI agents.
- CRM automation. Triggers actions from CRM data: a status change sends a message, creates a task or assigns an adviser. It acts on what is known and does not converse.
The agent works at the moment of the question; automation works on the record afterwards.
The decision matrix
| Job to be done | FAQ chatbot | AI admissions agent | CRM automation |
|---|---|---|---|
| Answer a requirements question at night | Only if anticipated | Yes, from published pages | No |
| Qualify an enquiry (program, province, domestic or international) | Fixed questions | Conversational, adapts | Only from captured data |
| Book an information session or call | Link to a form | Can offer times in the chat | Confirmation and reminders |
| Follow up after a session or campus visit | No | Not its role | Yes, its strength |
| Keep records consistent and deduplicated | No | Passes data to the CRM | Yes, with matching rules |
| Route an enquiry to the right adviser | Basic | Can summarize context | Yes, rules and queues |
| Handle a sensitive or complex case | No | Hands over to a person | Creates a task |
The agent is strongest before a record exists, automation after.
Who owns what
| Component | Natural owner | What they decide |
|---|---|---|
| Published answers the agent relies on | Recruitment with the registrar's office | Which pages and policies are the source of truth |
| Escalation rules to staff | Admissions manager | What triggers a handover, and the response time |
| CRM fields, statuses, workflows | Enrolment operations | Data model, triggers, sending rules |
| Consent and suppression | Privacy officer | Consent by channel, retention |
| Review of conversations | Recruitment with the supplier | Gaps to fix, wrong answers to correct |
Without owners, the tools contradict each other: the agent offers a call that no workflow schedules, or a sequence emails a person who just opted out in the chat.
When to choose one, and when to combine
- Repeat questions go unanswered outside office hours. Start with an agent on your website, grounded in published pages.
- Enquiries arrive but follow-up is slow. Start with CRM automation.
- Records are duplicated. Fix data first; both tools amplify bad data. See our guide to deduplicating student records across channels.
- Both gaps exist. Combine them, with the agent capturing a qualified enquiry and the CRM owning what follows.
Set the boundary in writing. The agent collects, qualifies and books, then hands the conversation summary and consent record to the CRM. If both tools message the same applicant, you have built a collision.
Privacy and messaging: PIPEDA, CASL and Law 25
PIPEDA. The Office of the Privacy Commissioner of Canada publishes guidance on PIPEDA for private-sector organizations. Public universities are often covered by provincial public-sector or privacy legislation instead, so confirm which regime applies to your institution. Either way, the principles are consistent: limit collection to what is needed, obtain meaningful consent and be open about how data is used.
CASL. Canada's Anti-Spam Legislation governs commercial electronic messages, and the CRTC enforces it, and the text of the Act is public. It matters most for automation: email and text sequences need consent that fits the rules, and every message needs identification and an unsubscribe mechanism. Whether a given message counts as commercial is a question for counsel.
Quebec, Law 25. Institutions operating in Quebec face the amended private-sector privacy law, overseen by the Commission d'accès à l'information, with resources on its website. It brings requirements on transparency about automated processing, privacy impact assessments in certain cases and rules on transfers of data outside Quebec. A hosted AI agent can trigger those provisions. Ask your supplier where data is hosted, and take advice.
For both tools. Tell people they are speaking with AI, set a transcript retention period and make suppression work across systems. Offer French and English service where your obligations or audience require it. The agent routes and informs; admission decisions stay with your staff and published criteria, including those used via OUAC or provincial application centres.
Canadian context: provincial systems and international applicants
There is no single national application system. Ontario universities use OUAC, other provinces run their own, and CEGEP students in Quebec follow a separate path. An agent that answers from your published pages can explain program-specific requirements and deadlines; automation structures the steps afterwards. Study permit questions come up often from international applicants. Answers should point to official government sources and never promise outcomes.
What to ask a supplier
- Which sources does the agent answer from, and what happens when it has none?
- How does it hand over to a person, and what does the adviser see?
- Where is data hosted, and is it used to train models?
- Which CRM objects does it read or write? Ask for it in writing.
- How do consent and unsubscribes pass between the agent and the CRM?
- Who reviews transcripts, and how often?
Measure your own baseline first: after-hours response time, follow-up speed, duplicate records. No public source provides a dependable benchmark for your institution.
Where Skolbot fits
Skolbot Chat is an AI agent installed on the institution's website. It answers from published content, qualifies the enquiry, offers an appointment or session and passes the conversation to your team and CRM. It supports advisers rather than replacing them. Workflows and reporting stay in your CRM; how the connector works with a specific CRM is confirmed during scoping. For the wider picture, read our guide to recruiting more students.
