Questions answered from the source
Staff and researchers get the answer from your own regulations, policies and service documentation, with the clause cited.
Education and research · AI agents inside the institution
Education and research agents. A person decides. Data stays.
Katonic runs AI agents inside your own data centre or private cloud. Agents read your regulations, your funder rules and your student records, prepare the work, and stop for a named person.
No agent admits or rejects an applicant, sets a mark, decides an appeal or a misconduct case, or submits a grant application. It prepares, and a person decides.
Student records and unpublished research stay on infrastructure you control.
Student records and unpublished research stay on your infrastructure
A model you serve
Each person's permissions kept
No agent decides about a student
Applications, claims, appeals, grant drafts and ethics forms run on reading, checking and drafting. That is where an agent helps most, and where a wrong decision about a person costs most.
The admissions templates we opened email applicants with no step at which a person approves.
None of the vendor pages we read offers on-premises hosting. And running there is not enough: what matters is which tool waits for a person, whose identity the agent uses, and what is recorded.
A decision about a student belongs to a person. In the European Union, AI used for admission and for evaluating learning outcomes is high-risk from 2 December 2027.
Tools must be chosen one by one: of 32 public education and research interface descriptions, the importer read 30, and those 30 carry 2,892 operations, 1,186 sent as changes.
Student records are personal data, and unpublished research is the institution's own. Both have to stay inside the institution.
What the rules ask
A person can oversee and stop the system. A person, not software alone, decides what significantly affects a student. No system infers a student's emotions. The institution knows where the data is.
Agents read and draft. People keep every decision about the applicant, the student and the research.
Questions answered from the source
Staff and researchers get the answer from your own regulations, policies and service documentation, with the clause cited.
Work prepared for a decision
Applications, claims, appeals, grant drafts and ethics forms are checked and set out for the person or the board that decides.
Patterns you can act on
Evaluation comments, timetable clashes and standards with no evidence are listed and counted, with the records behind each.
A record of every step
Who asked, which tool ran and what was decided.
From admissions and student services to the registry, teaching, research, the library, campus services, quality and finance.
Admissions and students
Enquiries, applications, funding checks, advising notes
Registry and teaching
Progression lists, claims, appeals, course outlines
Research and library
Grant calls, ethics forms, data plans, open access
Campus, quality and finance
Service desk, timetables, accreditation, invoices
Each is a use case a university or a research institute can build in Katonic Studio. A name with a link has its own page.
Every agent follows one rule: it prepares, and a named person decides.
Go deeper
Guides and answers
A university that wants AI agents on its own infrastructure has more than one kind of product to choose from. The products sold for this fall into four types: agent platforms you run yourself, infrastructure platforms, automation platforms with agent features, and developer frameworks. Our comparison describes fifteen of them with the same fields, read on each vendor’s own pages: where each runs, who operates it, governance, price and limits. For a university, a price that suits many occasional users and a free or open edition to start with are worth comparing first.
On-premises AI agent platforms compared: 15 options by type →An agent reaches your documents and systems in one of four ways, all through the same tool gateway and the same approval rules.
Connection · 01
An application, a transcript, a module outline, a grant call, an ethics form.
Connection · 02
Regulations, policies, programme handbooks, admissions criteria, funder rules and research procedures, indexed with each reader's permissions kept.
Connection · 03
↗Give Katonic the description of an interface you already run, and choose which operations become tools. This is the route to student records, courses and enrolments, the library catalogue, research outputs and the service desk.
Connection · 04
↗Servers from the catalogue, deployed inside your own cluster, for the data warehouse, reporting databases, research indexes and web search.
System names are left out on purpose. You supply the address and the sign-in. Nothing here is a tested integration with a named product.
It runs on the institution's own infrastructure or private cloud, with a model the institution serves. Student records and unpublished research stay there.
Every tool call passes the institution's policy. A policy can allow a tool, record its use, hold it for a person's approval, or block it, for the whole institution or for one department.
A tool that changes something waits for approval unless the institution has decided otherwise for that tool. That is how no record, reply or return goes out unapproved.
No agent decides about a student or an applicant. Admission, marks, progression, appeals and misconduct stay with the people and boards the regulations name.
The agent acts as the person using it. Where the institution's API checks the caller, Katonic passes the signed-in user's identity, and the agent cannot claim another.
Every call is recorded, with who asked, which tool, and what was decided, and an agent, a tool or a whole server can be switched off at once.
From visual workflows to full code - pick the path that fits your team and project.
Visual Builder
Follow guided steps to build agents. Configure each stage through the UI. No code required.
Best for
Registry, research office and quality analysts
Conversational Builder
Staff describe how they check an application for completeness or a grant draft against a funder's rules. AI generates the agent. Developers refine together.
Best for
Staff + Developers together
Full IDEs
JupyterLab, VS Code, RStudio with GPU access, pre-installed frameworks, and direct deployment.
Best for
Data Scientists & ML Engineers
The rules on admission, assessment, student data and research funding differ by country and point the same way. Each is a matter of how the system is built.
A person decides admission and marks
In the European Union, AI used for admission and for evaluating learning outcomes is high-risk from 2 December 2027. Here no agent admits, rejects or ranks an applicant, and no agent sets or proposes a mark.
No emotion inference, no proctoring
AI that infers emotions in education is banned in the European Union. Here no agent infers emotion from a student's words, voice or face, and no agent proctors a test.
A person, not software alone
Under the GDPR, a person, not software alone, decides what significantly affects a student. Here progression, appeals, misconduct and support stay with the people and boards your regulations name.
Funders restrict AI in research
Funders restrict AI in peer review and in writing applications. Here an agent checks a draft against the funder's rules and lists each breach. It writes no part of the case for support and no peer review.
Katonic AI is certified to ISO 27001. This is a summary of published rules, not legal advice.
In a school office, an administrator asks what the regulations say about an extension, and an agent answers with the clause cited; an exception goes to the policy owner. In the research office, an agent checks a draft grant application against the funder's rules and lists each breach for the researcher, who corrects, writes and submits. In the quality office, an agent maps the university's documents to each standard of a review and lists the standards with no evidence for the quality lead, who writes the submission. Each one reads what the university already has, changes nothing by itself, and leaves a record.
Seven questions to ask any supplier before an agent touches a student record, with our answers.
On infrastructure you control: your data centre or your private cloud. The model is one you serve there.
What the person using it can read. Knowledge sources keep each reader's permissions, and where your API checks the caller, Katonic passes the signed-in user's identity.
Only what you allow. A tool that changes a record, a reply or a return waits for approval unless you have decided otherwise for that tool.
A named person or the board your regulations name. No agent admits or rejects an applicant, sets or proposes a mark, or decides progression, an appeal or a misconduct case.
No. An agent drafts, and a member of staff reads and sends what goes to a student or an applicant. When the documents do not answer, the agent says so.
Who asked, which tool ran and what was decided, for every call.
At once. An agent, a tool or a whole server can be switched off by your administrators.
A student policy assistant can run on the first day, because no student record is touched. Talk to our team about building it on your infrastructure.