Comprehension data your security team can sign off on
alcoia gives an institution the same thing it gives a single instructor, across every course that uses it: a view of where readers struggle in assigned material, without collecting a record of what any individual read. It is deployed on terms your review can approve.
What your institution gets
Aggregate comprehension across courses, so departments can see which assigned readings consistently lose people and act on it. A site licence covers your seats. Instructors get the class view described on the educator page; the institution gets it at scale.
Anonymous by default
Instructors see grouped results, never individual students. Each student's outcomes are recorded under a pseudonym tied to a secret that is unique to one assignment, so the same student counts as one person within a reading but cannot be linked across readings. When that secret is deleted, on a retention schedule, the group figures survive and re-identification becomes impossible. Results drawn from fewer than five readers are withheld, so a small class cannot expose one person.
No fingerprinting
alcoia does not identify readers by device. No IP-derived identifier, no browser, canvas, font, or audio fingerprint. The free tier runs on a random token the reader can delete. This is a product rule, not a setting.
Where your material goes
Passage text is sent to a language model to generate questions and explanations, with nothing identifying attached. An institution can supply its own model endpoint, so that content goes only where you decide and never reaches our default provider. For institutions that require it, a self-hosted deployment keeps the whole system inside your own infrastructure. Both are scoped with you during procurement.
Self-hosted for complete data control
For institutions with strict data residency requirements, self-hosted deployment runs alcoia entirely within your infrastructure. Your student data never leaves your network. The extension client communicates with your own alcoia instance, which handles question generation and all data storage. This option is scoped with your team during procurement to match your security architecture.
What we can put in front of your reviewers
Model provider and hosting region, set with you. A sub-processor list naming every party that touches data, delivered with the data processing agreement. The client extension is open source under AGPL-3.0, so your team can read exactly what runs on a student's machine.
Students under 18
The product is used in education, including by students under 18. In the default anonymous mode there is no individual reading history on our servers, which reduces exposure under regimes such as COPPA and FERPA. We provide the documentation your review needs and work through your specific requirements; we do not ask you to take the claim on trust.