RSI separates a distance education course from a correspondence course, and the distinction carries Title IV funding with it. Reviewers ask for interaction that the instructor initiates, that is substantive, that happens regularly, and that you can show them.
Recorded lectures usually fail that test. A view count proves attendance at best. Annoto turns the same video into an interaction record: instructor prompts on the timeline, graded questions, threaded discussion tied to the moment it refers to, and per-learner participation that exports for review.
Federal regulation describes substantive interaction as engaging students in teaching, learning and assessment, and lists the kinds of activity that qualify: direct instruction, assessing or providing feedback on work, providing information about course content, and facilitating group discussion. It has to be initiated by the instructor, happen regularly and predictably, and be documented. Reviewers want records, not intentions.
Because watching is not interacting. A lecture posted to the LMS produces a view log, and a view log answers a different question from the one an auditor is asking. Programmes that lean on recorded content often carry their RSI burden entirely in discussion boards and email, which puts the load on instructors and leaves the video itself as dead weight in the evidence file.
An instructor placing a reflection point or a graded question at a specific moment is initiating interaction, and the timestamp records when. In-video quizzes authored by the instructor, prompts that open a discussion thread, and polls placed on the timeline all originate with the teacher rather than waiting for a student to ask.
Assessment and feedback on student work is explicitly named in the regulation. Peer review and video assignments put instructor feedback on a student's own performance footage. Threaded discussion anchored to a moment lets an instructor answer the specific confusion rather than post a generic announcement, which is the difference reviewers look for between substantive and administrative contact.
Predictable scheduling is part of the standard. Because every interaction in Annoto carries a timestamp and an author, analytics and CSV or API export produce a per-learner, per-week record of prompts issued, questions answered, threads replied to and work assessed. That is a documentation trail rather than a description of practice.
It does not have to. Most programmes keep the board and add the video layer, which spreads the interaction load across the content students were going to watch anyway. Faculty tend to notice the difference first: questions arrive attached to the minute that caused them, so answering one student often resolves the same confusion for the cohort.
Inside Canvas, Moodle, Blackboard or Brightspace, an admin registers Annoto as an LTI 1.1 or 1.3 tool, and sign-on, rosters and grade passback run through the standard so the records tie to real enrolments. On a portal or custom platform, a script tag and a client ID attach the layer through the JavaScript SDK, with a REST API for pulling the evidence into your reporting stack.
They fail separately, which is why a programme can pass one and miss the other. Regular is a property of the schedule: a pattern students can anticipate, stated in the syllabus and proportionate to the credit hours. Substantive is a property of the act. A weekly deadline reminder is regular but administrative; one long round of feedback in week eleven is substantive but not regular. Reviewers apply both at once.
The ones authored once and reused. A question set built over a lecture, with Lumo AI drafting from the transcript, takes minutes to edit rather than an hour to write, and travels with the video into next term and across parallel sections. In-video quizzes carry the assessment half; two timeline prompts per lecture carry the discussion half. What does not scale is asking faculty to originate fresh content weekly.
By checking it while the term runs. Attention and comprehension analytics break activity down by week and by learner, so the weeks where the record thins surface while there is time to correct them. Check per section, not per course: a course taught in six sections needs each instructor of record visible in that section's own data, not one well-run section carrying the programme.
Four things per interaction: who authored it, when, which video and moment it attaches to, and what kind of act it was. Graded items reconcile through LTI 1.3 gradebook sync, so the LMS score and the interaction behind it are one record, not two accounts a reviewer aligns by hand. Pull the CSVs at term close, filed with the syllabus and course shell.
Four findings recur. Interaction that is student-to-student with the instructor absent. Contact that is real but administrative: reminders, grade notices, availability. A cadence that holds in one section and nowhere else. And an evidence file assembled after the request, which reads as reconstruction. Video assignments and peer review answer the first, provided the instructor's own assessment sits alongside the students'.
Same lecture, two very different evidence files.
The four tests a reviewer applies, and the artefact that answers each one.
Annoto produces the interaction record. Whether a given course meets RSI remains a judgement for your institution and its accreditor, and this page is not legal advice.
Start with the courses carrying the most recorded content and the least documented interaction.
An LMS admin registers Annoto over LTI 1.1 or 1.3, or you add a script tag on a portal. Choose the online programmes under review first.
Agree a pattern faculty can keep, such as two instructor prompts and one graded question per lecture, so the record is regular rather than sporadic.
Pull per-learner participation by CSV or API into the evidence file, before a reviewer asks rather than after.
Where the interaction record comes from, and what else it satisfies.
Twenty minutes. Bring an online course that leans on recorded lectures, and we will show you what the export looks like.
Book A Demo