One workspace for writing centres, integrity offices, admissions, libraries, and IT, with SSO, FERPA-aware data handling, and the same sentence-level evidence faculty already trust in their personal accounts. Calibrated for academic prose so ESL writers and methods-heavy STEM submissions do not over-flag. Volume pricing, signed DPAs, and a 30-day institutional pilot if procurement needs proof before the multi-year contract.
Universities reach for AI detection in at least four places at once. A shared institutional workspace replaces four separate logins, four billing lines, and four versions of "which detector are we using this semester".
The standard pattern in 2026 is the same tool serving the writing centre, the graduate-programme office, the academic integrity office, and the admissions team, with separate seat groups and shared billing. Each group keeps its own scan history, integrity boards see the audit trail they need for review hearings, and IT has one vendor to manage instead of five.
Tutors use detection during student consultations to anchor revision conversations on specific paragraphs rather than a vague sense that the prose reads AI. Sentence-level highlights make the conversation concrete, and the audit log captures who reviewed what for departmental reporting.
Theses, dissertation chapters, and qualifying exams go through committee review. Bulk PDF upload covers a full cohort of submissions in one workspace, scan history retains every chapter pass through the defence, and saved PDFs anchor the file if an integrity question arises post-defence.
Integrity coordinators run scans alongside the standard Turnitin or iThenticate output. The structured audit log, sentence-level reasoning, and exported PDF reports become defensible evidence for a hearing, not a screenshot pasted into an email thread.
Application essays, personal statements, and writing samples flow through the REST API into Slate, Common App back-end systems, or a custom admissions portal. Decisions stay with the human reviewer; the score and highlights inform the read without replacing it.
A percentage on a gradebook does not stand up in an integrity hearing. Universities need evidence that a reviewer can read aloud to a panel and that a candidate can respond to.
Every submission produces an Authenticity Score and a sentence-by-sentence colour map showing which passages cluster red. Perplexity, burstiness, and sentence-variance signals are exposed individually, so a reviewer can see why a passage flagged rather than only that it did. The unit of action is the sentence, which is the unit a faculty member, an integrity coordinator, or a writing-centre tutor can have an actual conversation about.
International student populations and STEM disciplines see higher baseline AI scores because structured, rule-following prose overlaps with model output. The classifier is calibrated against academic corpora, and the dashboard surfaces the discipline context so reviewers calibrate rather than react to a single number. The result is a lower false-positive rate on exactly the writing universities serve most.
Student essays, dissertations, and application materials submitted through your workspace never enter our training pipeline. This is a contractual commitment in the DPA, not a marketing line, and it survives renegotiation when your legal team needs to mirror the language to an existing data agreement.
EU institutions get a standard GDPR data-processing addendum with sub-processor disclosure and deletion-on-request. US institutions get DPAs that mirror most FERPA-aligned vendor expectations. We do not yet hold a formal FERPA certification; the architecture is privacy-aware by design and we walk through vendor risk assessments on request.
Business is the starting institutional tier for a department or single office. Faculty-wide and multi-campus deployments move into Enterprise. Full details on the pricing page.
Billed $89.88/year — Save $30
Billed $179.88/year — Save $60
Billed $359.88/year — Save $120
Faculty-wide and multi-campus deployments scale into Enterprise with SSO, additional seats, and custom DPAs. Talk to sales →
The pattern that works across hundreds of universities is to ship in one department first, prove the workflow, and expand on the renewal cycle rather than try to land the whole faculty in a single contract.
The writing centre, the graduate-programme office, or the academic integrity office takes a Business workspace with five seats. The director, two senior tutors, and a coordinator share one billing contact. Most departments are productive inside the first week because the UX is the same one their faculty already use in personal accounts.
Once one team is running steady, adjacent departments either join the existing workspace as additional seat groups or take their own Business workspace under the same parent contract. Cost-centre tags on Enterprise let each department see its own usage line in the invoice without splitting the billing relationship.
At the faculty level the deployment moves into Enterprise, which adds SSO, larger seat counts, the institutional audit log, and a single procurement relationship. Most universities reach this stage in semester two or three of adoption, after the workflow has proven itself inside the early departments.
The pilot programme exists to give procurement a clean evaluation window before the multi-year contract. Standard institutional pilot is 30 days, 10 named reviewers, prorated at the standard Business or Enterprise rate.
When an integrity hearing asks "why did you accuse this student", the answer needs structure. Every scan logs a timestamp, the reviewer, the input hash, the model version, and the sentence-level reasoning.
Each scan is retained in the institutional workspace with the original input, the score, the sentence-level breakdown, and the model version that produced the score. Integrity coordinators can pull the record for a specific student or a specific assignment without asking the original reviewer to search their email. Retention windows are configurable per workspace and follow institutional policy.
Two reviewers scoring the same submission should see the same evidence. The shared workspace gives the integrity office a way to calibrate by running test scans across the team, surfacing scoring drift early, and writing institutional guidance on how to read the score before a hearing rather than during one.
The PDF export carries the AI score, sentence breakdown, model version, and timestamp in a consistent format. Board chairs see the same report shape in case 1 as in case 200, which keeps hearings focused on the evidence rather than on the layout of the document. White-label branding on Business and Enterprise lets the report carry institutional identifiers if required.
The score is a signal, not a verdict. Detection plus the conversation with the student, plus the institutional Turnitin or iThenticate record, plus faculty judgement, is what produces a defensible call. We tell integrity offices this directly in onboarding so the tool is positioned correctly inside the institutional process.
Most US R1, UK Russell Group, and EU research universities wrote formal AI policies between 2024 and 2026. The standard shape is documented evidence, a faculty conversation, and human judgement before any referral. The scan, the sentence-level breakdown, the saved PDF, and the institutional audit log together form the documented evidence layer your policy requires.
Departments differ on where the threshold sits. Some treat anything above a 70 percent AI score as a flag, others use 50, and some prefer to read the sentence-level evidence without a single-number cutoff. The dashboard supports all three patterns, and the institutional onboarding session walks through which approach fits your existing policy.
Turnitin and iThenticate stay as the official submission-layer record. TextSight runs alongside as the sentence-level evidence tool. The two are complementary, not duplicate, and most institutional deployments keep Turnitin as the gradebook check and TextSight as the integrity-office working tool.
University IT carries the security review. The posture is built so the vendor risk assessment reads as routine, not as a six-month negotiation.
Reviewers sign in with their university identity through Okta, Azure AD, Google Workspace, Shibboleth gateways, or whichever federation the institution runs. Provisioning and deprovisioning follow the identity provider, so a leaver loses access on the same day their email is suspended. Business teams sign in with email and password plus optional two-factor authentication; if procurement requires SSO at the Business price point we scope it case by case.
Every scan logs a timestamp, the operator, the input hash, the model version, and the sentence-level reasoning. The audit log is exportable from Enterprise workspaces and is suitable for academic integrity review boards that need a consistent record across multiple graders. IT can review the log for unusual usage patterns the same way they review any institutional SaaS.
Student essays, dissertations, and application materials submitted through your workspace never enter our training pipeline. The clause is in the standard DPA and survives renegotiation when the legal team needs to mirror an existing institutional data agreement. This is the question that decides the procurement on most security reviews.
Default infrastructure is US-hosted. EU data residency is on the 2026 roadmap with delivery scoped for institutional Enterprise customers that confirm volume in advance, and UK and Canadian residency follow shortly after. We will be transparent about the timeline rather than promise residency we cannot deliver this quarter.
All data encrypted in transit with TLS 1.3 and at rest with AES-256. Backups carry the same encryption profile. Seat-level permissions control who can see whose scan history inside the workspace, and sub-processor disclosure is provided in the DPA on request.
Site licences, signed DPAs, vendor risk assessments, and invoice terms that match how universities actually buy.
Business is the starting institutional tier at $39.99 per month per workspace. Faculty-wide and multi-campus deployments move into Enterprise where seat count, API quota, and contract length are negotiated case by case. Annual prepayment qualifies for the 25 percent annual discount, and multi-year contracts are available on Enterprise with a price-hold clause across the contract term.
US institutions get a standard DPA that mirrors most FERPA-aligned vendor expectations. EU institutions get a GDPR data-processing addendum with sub-processor disclosure, deletion-on-request, and standard contractual clauses where required. UK and Canadian institutions are covered under equivalent terms. The legal team will negotiate language to match an existing institutional data agreement where required.
Standard vendor onboarding paperwork, W-9 or equivalent tax forms, and security questionnaires are completed on request. Most US, UK, and EU institutions clear vendor review on the standard DPA without renegotiation. Where renegotiation is needed, the typical turnaround is two to three weeks with named legal counterparts on both sides.
We accept standard purchase orders, invoice-based payment with net 30 or net 45 terms, and direct wire transfers for institutions that cannot pay by card. Split invoicing across cost centres is available on Enterprise contracts where each department wants its own invoice line for internal recharge. Reach out to enterprise@textsight.ai with seat count, billing entity, and procurement timeline.
Two and three-year contracts qualify for additional discounting beyond the annual rate and include a price-hold clause for the contract duration. Renewal is on the anniversary; cancellation requires standard notice and is honoured without penalty.
The individual faculty workflow for graduate seminars and pre-submission journal scans.
Faculty workflow →The student-side workflow for pre-Turnitin scans and false-positive defence.
Student workflow →REST API for LMS and admissions integration. Reference implementations for Canvas in progress.
Read the docs →How the classifier is tested, where false positives concentrate, and how to read the signals.
See the methodology →Talk to sales about SSO, signed DPAs, the 30-day institutional pilot, and volume pricing for faculty-wide deployment.