Conversations about AI in higher education tend to swing between two poles: tools that will do everything, and rules that forbid anything. Most faculty we speak to want something more practical: AI that takes the first draft of repetitive work and leaves academic judgment with them. AcademicOS, the product family from the team behind Proctorly, is built on one rule that captures this: AI drafts, people decide.
Below, we go task by task through where we think approval has to sit.
Course design: approve before anything is published
Turning a syllabus into modules, lessons, activities and questions takes most of the effort in building a course. A structured first draft saves real time, as long as it follows your syllabus and outcomes and a qualified person checks it. In Studio, our AI course creator, every drafted item is mapped to a course outcome and a Bloom’s level and goes through an approval workflow. Nothing is published to learners until someone signs it off.
The approval step also gives a dean a view of how a course was built, what has been reviewed and what is still a draft.
Teaching preparation: suggestions, not instructions
During a term, the AI Teaching Copilot suggests lesson plans, activities and examples aligned to each session’s outcomes, and drafts questions and rubrics at a set difficulty. Faculty accept, edit or discard each suggestion. Read more about the AI-powered teaching assistant.
This approval is lighter than for a published course, and it still matters. A rubric shapes how students are judged, and a question at the wrong level skews an assessment. A faculty member who knows the cohort is the right person to make that call.
Feedback on submissions: the clearest line
Draft feedback is one of the most tempting uses of AI, because it is where faculty time runs out. It is also where an unchecked draft can do the most damage, since a student reads feedback as their teacher’s judgment. In AcademicOS, suggested feedback is never sent unless a faculty member has accepted or edited it. We think every institution should hold that line, whatever tools it uses.
Learner questions: answer from your documents, hand off decisions
Students ask the same questions every week: deadlines, policies, who approves an extension. A support assistant can answer the routine ones at any hour, provided it answers only from the institution’s own handbooks and shows where each answer came from. AskOS, our chatbot for education, works that way, and hands requests that need a decision, such as an extension, to the right member of staff.
The approval here happens upstream. Staff decide which documents AskOS can draw on, and they keep every decision the documents do not settle.
Early notice: bring a learner to a person’s attention
Progress data can show who is falling behind while there is still time to help. The Copilot gives faculty early notice. It does not act on a learner by itself, and it does not guess at a learner’s motives or state of mind. What happens next is a teaching decision.
Five things AI should not do alone
- Publish content to learners without a person’s approval.
- Send feedback that faculty have not accepted or edited.
- Grade a learner or decide an academic outcome.
- Answer from sources the institution did not provide, or make a decision that needs a member of staff.
- Infer a learner’s emotion, character or intent.
The same principle runs through Proctorly’s exam products. AI-drafted questions enter the question bank only after a moderator approves them, and AI mark suggestions in on-screen marking leave the evaluator with the final say.
Approval as evidence
Building approval into the workflow, instead of leaving it to good intentions, has another benefit. Because lessons, questions and rubrics are mapped to outcomes and approved on record, accreditation evidence builds up as faculty teach. When a review committee asks how a course was designed and assessed, the answer is already in the system.
Frequently asked questions
Do we need Proctorly to use AcademicOS?
No. AcademicOS is a separate product family. The two connect when a course moves to formal assessment, but each works on its own.
Can we start with one AcademicOS product?
Yes. Studio, the AI Teaching Copilot, AskOS and the AcademicOS LMS share sign-in and data, so many institutions start with one and add others later.
Want to see AcademicOS on your own syllabus? Book a demo.