Campus-wide detection that your writing center won't fight.

About: Higher education

Higher-education deployments of AI-detection tools fail in a specific way. The academic-integrity office buys the tool. The writing center finds out from a student who was wrongly flagged. The faculty senate passes a resolution asking who approved this. Procurement gets a letter from the faculty union. The tool becomes a political object.

We've seen this happen and we think it's largely avoidable. What avoids it: transparency about the tool's limits, a pilot that includes the writing center from day one, an appeal procedure that doesn't route through a black box, and a pricing structure that doesn't incentivize aggressive use. This page is our pitch for how that deployment looks.

What universities get

• LMS integration. CSV round-trip with Canvas, Blackboard, Moodle, and D2L Brightspace today; native LTI 1.3 integration in fall 2026, institutional plans first.
• SSO. SAML 2.0 (Shibboleth, Okta, Entra ID, Google) or OIDC. Students never see our login screen.
• Retention policy. Configurable per institution; the default for higher-ed is 30 days to match our standard, but institutions can dial to zero (scan-and-discard) if that's what the campus IRB or legal counsel requires.
• Published methodology and false-positive rate. On /transparency, updated quarterly, with the benchmark datasets named. This is what the faculty senate will ask for.
• GDPR alignment for European students and study-abroad programs. Data Processing Agreement, subprocessor list, right-to-erasure workflow.
• Writing-center partnership track. We offer a no-cost writing-center license alongside the campus license so the writing center can review detector reports as a support function, not as a gatekeeper.

Pilot design we recommend

1. One college/department, one semester. Ideally a writing-heavy department (English, History, Communications) that's willing to pilot.
2. Include the writing center. Writing-center staff review the flagged essays alongside instructors. This catches false-flag patterns early and builds trust.
3. Publish the false-positive rate from the pilot to the faculty senate and academic-integrity committee. Transparency pre-empts most of the political friction.
4. Iterate on the appeal procedure before broad rollout. The appeal procedure is the thing that determines whether the tool is a support or a surveillance tool.

What we don't do

We don't integrate with plagiarism detection as a combined score. We think merging AI-likelihood with text-match-to-web is a category error, they measure different things and blending them produces false confidence. We also don't provide student-level analytics, per-student scoring histories visible to administrators, or any form of "risk scoring" beyond the per-essay probability.

Pricing

Quoted per enrolled-student FTE, with academic-year and multi-year pricing. Like K-12, we don't list public higher-ed pricing because it depends on the LMS integration scope and the retention policy chosen. Request a quote via /contact.