TurtleDock
CampusHub · Feature

Screening applications faster, without deciding them

CampusHub uses automated screening to order and flag admission applications: checking document completeness, extracting marks from uploaded transcripts for verification, identifying duplicates, and ranking against your published criteria. Every admission decision remains with the institution.

It does not decide admissions and does not score candidates on anything beyond your published criteria. Extracted values are always shown for human confirmation before use.

Screening three thousand applications is a document-checking exercise before it is a judgement. Roughly nine in ten of the hours go into confirming that a mark sheet says what the form claims, that nothing is missing, and that the same candidate has not applied twice.

Does
Completeness, extraction, duplicates, ranking
Does not
Decide, reject, or score beyond your criteria
Extraction
Shown for confirmation, never applied silently
Audit
Every flag recorded with its reason
01

Where automation is appropriate here, and where it is not

The distinction is between checking and judging.

TaskAutomatedWhy
Document completenessYesA checklist, with no judgement involved
Reading marks from a transcriptYes, for confirmationExtraction is fast and error-prone, so a human confirms before it is used
Duplicate detectionYesSame candidate under two applications is a factual match, not a judgement
Ranking on published criteriaYesDeterministic arithmetic on criteria you published in advance
Assessing suitabilityNoNot defensible, not lawful to delegate, and not something we will build
02

Why extraction is always confirmed

Reading a mark from a scanned transcript is exactly the kind of task where automation is right most of the time and wrong in ways that are hard to spot — a smudged digit, an unusual layout, a subject listed twice.

Extracted values are presented alongside the source image for a human to confirm. That is slower than accepting them silently, and it is the difference between a merit list you can defend and one you cannot.

FAQ

Questions people ask

Phrased the way they arrive, answered so each one stands on its own.

Does AI decide who gets admitted?

No. It orders and flags applications against criteria you publish, and every decision is made by the institution. Delegating an admission decision to a model is neither defensible nor something we build.

How accurate is mark extraction?

Good enough to save time, not good enough to trust silently. Every extracted value is shown beside the source document for confirmation before it enters the merit calculation.

Can it detect duplicate applications?

Yes, across name, date of birth and identifiers, flagged for review rather than merged automatically. Two genuinely different candidates occasionally look identical on paper.

Related

Read next

Provenance

Who checked this, and against what

Accountability

Checked by the Implementations desk on 1 September 2026. Corrections go in the page, dated, rather than quietly.

No call required

Tell us your enrolment, we will tell you the number.

Pricing is on the page, in the open, with the counting rule written next to it. Nobody has to book a call to find out what it costs.

See pricing

Bring your own structure. We configure it live on the call.

Forty minutes with an implementation engineer — your classes, your fee heads, your grading scheme, entered as we go. No slides, no sample school.