Solution

A knowledge base for your institution's handbooks and procedures

An AI knowledge base for education answers the administrative and policy questions that staff and students ask an institution every day — what the handbook actually said the year a student enrolled, whether a module's prerequisites allow a particular sequence, what the internal process is for a credit transfer or a leave of absence, what an accreditation body's documentation requires. It retrieves the institution's own written material and cites the source document and its version on every answer, so the person reading it can check the original before acting on it.

· Reviewed by Artur Horimoto, Founder & CEO

The policy that changed since anyone last read it

A school or university runs on documents that are supposed to be authoritative and almost never are, in practice, the moment someone actually needs one. The student handbook gets revised most years — extenuating circumstances rules, attendance requirements, academic integrity procedures, appeals processes — and each cohort is technically governed by whatever version was in force when they started, not necessarily the version currently on the website. A registry officer fielding an appeal from a student who enrolled three years ago needs the handbook that applied that year, not this year's, and getting that wrong is not a small error — it can mean applying a rule that never governed the case in front of them.

The same problem repeats everywhere written policy meets a real decision. A module catalogue lists prerequisites and credit values, but catalogues are revised between intakes too, and a student asking whether an old module still satisfies a current prerequisite chain needs the version that matches their year, not the live one. Internal process documentation — how a professional services team handles a credit transfer, a deferral, a data request — sits in procedure manuals that get updated piecemeal, so the copy someone remembers from training may already be out of date. Course teams preparing for an accreditation visit or an annual review need the exact wording of a regulator's or professional body's requirement, not a paraphrase from the last review cycle. And IT and facilities procedures — how a room-booking exception gets approved, what the process is for provisioning system access, who owns a broken piece of AV equipment — exist in writing somewhere but rarely where the person who needs them thinks to look.

None of this is a teaching problem. It is a findability problem with a specific cost: staff spend real time re-reading documents to confirm which version applies, or worse, answer from memory and get the version wrong.

What this is not

This is not a tutoring tool, and it does not do a student's academic work. It does not answer subject questions, does not explain course content, does not write, check, or improve an assignment, and it will not engage with anything that looks like a request for academic help. What it does is retrieve the institution's own administrative and policy material — handbooks, catalogues, process documents, regulatory and quality material, IT and facilities procedures — and hand back the passage that answers the question, with the source document and its specific version cited. If a question falls outside what is indexed, or reads as an academic request rather than an administrative one, the system says so rather than attempting an answer it has no basis for.

What the system does, day to day

Once it is built around your institution's actual documents, the mechanics stay the same regardless of who is asking or what they're asking about. The system searches the indexed material, finds the passage that actually answers the question, and returns an answer built from that passage — with the source document and its version cited every time. A handbook answer names which year's edition it came from. A catalogue answer names which academic year's module listing it pulled from. Nobody needs to know which shared drive a document lives on or what the file is called; the question gets asked in plain language, the way it would be asked of a colleague who happened to have read everything.

The material itself spans several distinct kinds of documents, and the system is built to hold all of them without blurring which is which: staff and student handbooks, where the applicable version depends on the reader's cohort or the year in question; module and programme catalogues, including prerequisite chains and credit rules; internal process documentation used by professional services teams — registry, admissions support, student finance, exams administration; regulation and quality assurance material that course teams reference for accreditation and periodic review; and IT and facilities procedures covering system access, room bookings, and equipment support.

Access mirrors whatever permission structure your institution already runs. Staff-only policy — an internal process document, a disciplinary procedure, material scoped to a specific department — does not surface to a student asking a general question, the same way it would not be handed across a desk to someone without the right role. A student's own handbook and catalogue material is appropriate to surface to that student; an internal registry procedure for handling their case is not. Every retrieval is logged against the person who asked, so there is a record of who looked up what and when.

Three ways this shows up on a normal week

A handbook question that depends on the year

A registry officer is preparing a response to an academic appeal from a student who enrolled two years before the current handbook took effect. The extenuating circumstances procedure has changed since then — the evidence window is different, and so is the approval route. Rather than working from the current handbook on the intranet, she asks the system what applied for that student's cohort. It returns the relevant section from that year's edition, cited by document and version, alongside a note that the current edition differs. She drafts the response against the rule that actually governed the case, instead of the one that happens to be live today.

A prerequisite question mid-advising session

A professional services adviser is sitting with a transfer student trying to work out whether a module completed at their previous institution satisfies the prerequisite chain for a course they want to take next term, and whether the credit value carries across cleanly. She asks the system, which pulls the relevant module descriptors from the programme catalogue — prerequisites, credit weighting, the internal process for logging an external credit transfer once it's approved — all cited to the current catalogue and the credit-transfer procedure document. The student leaves the meeting with an answer instead of a promise to follow up once someone checks.

A quality check before an accreditation visit

A programme lead is assembling evidence ahead of an external accreditation panel and needs the exact wording of a requirement referenced in the institution's own quality assurance file, not a summary someone wrote from memory during the last review cycle. She asks the system, which returns the specific clause from the quality documentation, cited to the source file and the version submitted at the last review point. She can quote it directly in the panel evidence instead of re-reading the whole document to confirm she has the current wording.

How it connects to what your institution already runs

The system indexes documents your institution already has rather than asking anyone to re-enter anything. Handbooks, module and programme catalogues, professional services process manuals, quality assurance and accreditation files, and IT and facilities procedure documents are all reasonable sources, wherever they currently live — a shared drive, SharePoint, an intranet, or a folder that has been meaning to move to a shared drive for years. Most student information systems and learning management platforms also expose an API or a scheduled export that can feed structured catalogue data into the same index, alongside the voice AI and AI agents Calfy builds for the enquiry and document-chasing side of the same institutions — this system answers the "what does the policy say" question those two don't.

Because handbooks and process documents change on a schedule — most annually, some more often — the index is built to hold every version it has seen, not just overwrite the old one, so an answer can be traced to the edition that actually applied at a given point in time. When a document is formally retired and replaced, the old version stays queryable as history rather than being surfaced as current.

Student data deserves the same care here as anywhere else it appears. Any material that touches student records is scoped to the narrowest access that lets the system do its job, with detailed logging of who asked what and when, and the specific obligations around how that data is collected, stored, and shared are ultimately defined by your institution's own data protection and compliance lead, not by us — we build to support that governance rather than claim to replace it.

This system is one piece of the knowledge systems Calfy builds for document-heavy, permission-sensitive parts of a business, and the use case for searching an internal knowledge base walks through the general mechanics if some of this is new — the glossary entry on AI knowledge bases is a good starting point for the underlying terms. For where this fits alongside everything else Calfy builds for schools and universities, the industry overview for education covers the full range.

Frequently asked questions

Is this a tutoring or teaching tool?

No. It does not answer academic questions, explain course content, or produce or improve student work of any kind, and it will not engage with a request that reads as one. It retrieves the institution's own administrative and policy material — handbooks, catalogues, process documents — and cites the source and version on every answer.

How does it handle policies that change every year?

Every handbook, catalogue, or process document answer names the specific document and version it came from, and the index retains prior versions rather than discarding them, so a question about a rule that applied to an earlier cohort can be answered against the edition that actually governed it, not the current one.

Will staff-only material ever surface to a student?

No. Access mirrors your institution's existing permission structure. A student asking a question sees student-facing material — their own handbook, the public catalogue — and an internal process document scoped to professional services staff does not surface outside that group, the same way it wouldn't be handed across a desk to someone without that access.

Will it work with the systems we already use?

In most cases, yes. Shared drives, SharePoint, intranets, and exports from a student information system or learning management platform are all reasonable sources. Bring the list of where your handbooks, catalogues, and process documents actually live to the first call — that's usually enough to scope what's realistic.

How much does this cost and how long does it take to go live?

It depends on how many document sources it connects to and how many distinct permission groups it needs to respect. We scope every engagement and agree clear pricing before build work starts, and most systems are answering real questions within weeks, starting with whichever document set is causing the most repeat questions first.

Bring the handbook version nobody can confirm, or the catalogue rule your professional services team keeps re-checking by hand. A free 30-minute strategy call is enough to tell you whether a knowledge system fixes it, and roughly what that would take.

See what this looks like for your business

Thirty minutes, your actual workflow, and a straight answer on whether this is worth building.

Free 30 minutes. No pitch deck. You leave with a plan either way.