A direct Blackboard Learn and Blackboard Ultra integration is in our planning queue, targeted for Q4 2026 and gated on waitlist demand. Until it ships, faculty can already use TextSight on Blackboard submissions via paste or DOCX upload. This page is the honest status update: what we are planning, what works today, and how to push your institution up the queue.
No vapourware. Here is the honest current state of the integration, the scope we plan to ship, and the demand signal we need to greenlight the build.
The Blackboard integration is currently in planning. We have not started the LTI 1.3 build. We have a design sketch, a faculty-research call list, and a list of waitlist signups that we read every Monday. Target ship window is Q4 2026, subject to demand signal. If waitlist density stays light, the build slips to 2027 in favour of higher-volume LMS requests. We will not pretend otherwise.
If you need any of the above before Q4 2026, the honest answer is the Blackboard integration may not be the right fit yet. We would rather tell you that now than later.
Specific to Blackboard, not generic LMS copy. These came out of conversations with faculty running Learn SaaS and Ultra at four institutions in 2026.
Today the workflow looks like: open submission in Blackboard, copy text or download DOCX, switch tab, paste into a detector, read the result, switch back, type a comment. For a class of 80 essays that is 80 context switches. The integration moves the signal into the Grade Center where the rest of the grading happens.
SafeAssign is excellent at text-matching against its index. It is not a transformer-based AI detector. Faculty who teach in writing-heavy fields keep asking for a second signal that addresses the LLM question specifically. The integration adds that without removing SafeAssign.
The Original Course View and Ultra render submissions differently. Faculty mid-migration tell us they want the AI signal to look identical in both views so they do not have to retrain TAs on two workflows. The plan is to ship Ultra and Original in the same release rather than staggering them.
Blackboard customers in the US, UK, Canada, Australia, India, and ESL-heavy programmes have flagged that generic AI detectors over-flag formally-taught English. TextSight's internal ESL false-positive rate sits at roughly 6 percent against a competitor band of 14-22 percent. The integration carries that calibration into Blackboard rather than installing a different model.
When a flagged submission goes to a misconduct panel, faculty need a defensible artifact: who scanned the essay, when, with what confidence score, against what model version. The Business-tier audit log gives that today via the web app; the planned integration writes the same log entry from inside Blackboard so the audit trail does not split across two systems.
Most faculty using TextSight inside a Blackboard course are running one of the three patterns below. None requires the integration. All are FERPA-compatible when you scope the data correctly.
Open the Grade Center, open the submission, copy the essay text, paste it into the TextSight web app, scan, read the sentence-level highlights, copy any flagged-sentence rationale into your Blackboard feedback comment. Works on the free tier (3 scans per day at 5,000 characters per scan) for low-volume spot checks. Pro at $19.99 monthly (or $14.99 monthly on annual billing) lifts the daily cap. .edu Pro is $13.99 monthly with a verified .edu email.
From the Grade Center, download all submissions as DOCX or PDF using Blackboard's batch download. Drop the folder into TextSight's Bulk Scan on the Business tier ($39.99 monthly, or $29.99 on annual billing). The Business tier ships up to 500 files per batch, full DOCX/PDF/TXT/MD extraction via officeparser v7, and a CSV export with one row per submission. The CSV is your defensible artifact for an academic-integrity panel.
Some faculty point students at the TextSight web app and ask them to pre-scan their own draft before submitting to Blackboard. This shifts the AI conversation upstream and reduces the number of post-submission flags. It also gives ESL students a chance to see what is tripping detectors and edit the rhythm, not the meaning. The free tier covers this for most students.
A short feature matrix so faculty making a 2026 vs 2027 decision can see what they get today, what they get when the integration ships, and what we still do not plan to ship even after launch.
| Capability | Today (paste workflow) | Planned integration | Not in v1 scope |
|---|---|---|---|
| AI-likelihood score on essay | Yes (paste / DOCX upload) | Yes (inline in Grade Center) | n/a |
| Sentence-level highlights | Yes | Yes | n/a |
| Per-sentence "why-flagged" rationale | Yes | Yes | n/a |
| Bulk DOCX / PDF scan | Business tier, 500 files/batch | Yes, course-level | n/a |
| Co-existence with SafeAssign | n/a (outside Blackboard) | Yes, side-by-side | Deep SafeAssign data exchange |
| Blackboard Learn SaaS support | n/a | Planned v1 | n/a |
| Blackboard Ultra support | n/a | Planned v1 | n/a |
| Original Course View support | n/a | Planned v1 | n/a |
| LTI 1.3 install at institution level | n/a | Planned | n/a |
| SSO via Blackboard auth | n/a | No (TextSight login) | Out of v1 scope |
| Automated gradebook column writes | No | No (manual grading) | Out of v1 scope |
| Mobile Blackboard app surface | No | No (web only) | Out of v1 scope |
| Audit log for misconduct panels | Yes (Business) | Yes, in-integration | n/a |
| FERPA-aligned data handling | Yes (in-memory, no model training) | Yes (inherited) | n/a |
| Estimated ship window | Live today | Q4 2026 target, demand-gated | n/a |
The "Planned integration" column is target scope, not a contractual commitment. We will update this page each quarter as build status moves.
We prioritise LMS integrations by waitlist density per institution. Tell us your campus name, Blackboard version, and rough class volume. Waitlist signups get a 90-day free trial of the integration at launch.
Today's signal path: email support@textsight.ai with subject line Blackboard waitlist and include:
We read every Monday. No marketing drip. We reply within 5 business days with current queue position.
Blackboard integrations touch student writing, which is FERPA-protected in US higher education. Here is the data-handling posture we operate under today, and what the integration will inherit when it ships.
Scan content is processed in memory to produce the AI-likelihood score and sentence-level highlights. Content is not used to train the detection model, ever. Retention windows are customer-controlled on the Business tier; the default is short-window retention only for the audit log and the user's own scan history. Customers who require a zero-retention deployment can opt for the no-storage mode at signup; in that mode, only the score and the audit-log entry persist, not the source text.
FERPA does not regulate vendors directly; it regulates the institution. Vendors operating on FERPA-protected data sit under the institution's "school official" designation or a written data agreement that pins down: scope of access, retention, breach response, data return / destruction at contract end, and prohibited secondary use. We sign that written agreement on request. The Blackboard integration when it ships will operate under the same terms, with the additional safeguard that LTI 1.3 launches only carry the specific submission the instructor opens (not the whole gradebook).
Before deploying the integration at your institution, your IT or compliance office will typically need: (1) a data processing agreement, (2) our subprocessor list, (3) a written breach-notification timeline, (4) confirmation of where scan content is processed (region), and (5) the no-model-training attestation in writing. We provide all five on request to support@textsight.ai.
We will not sell scan content. We will not use it to train competing models. We will not retain it beyond the customer-set retention window. We will not surface it to other customers. None of those will change when the integration ships. They are floor commitments, not features.
The hub page: detection workflow, ESL calibration, classroom-fit pricing, and faculty-friendly resources.
Visit the hub →Day-to-day classroom workflow patterns, score-band reading, and the appeal-process language we recommend.
Read the guide →Canvas LMS integration status, today's SpeedGrader workflow, and the waitlist signal.
See Canvas plan →Moodle integration plan, today's assignment workflow, and waitlist.
See Moodle plan →Google Classroom K-12 integration plan and Drive-native submission workflow.
See Google Classroom plan →Free $0, Starter $9.99 ($7.49 yr), Pro $19.99 ($14.99 yr, $13.99 .edu), Business $39.99 ($29.99 yr).
See pricing →Waitlist signups shape the order. Email support@textsight.ai with subject "Blackboard waitlist" and your institution name. Faculty on the list get a 90-day free trial at launch.