Regular and substantive interaction, read post by post in every online course, cited weekly and fixed mid-term
Every online section’s instructor interaction cited weekly, and gaps fixed mid-term.
See one case, screen by screen ↓Regular and Substantive Interaction Monitoring is software for online learning offices that checks every online course for regular and substantive interaction between instructors and students. Seven agents read every instructor post, reply, feedback comment, message and live session each night, label each one against the five federal elements with the deciding passage quoted, and measure gaps against the course's own posted schedule. When a gap passes seven days, a private, supportive note is drafted for the compliance director to edit and send.
AI in education: a buyer's guide to agentic AI for colleges, universities and school districts →
Updated
Regular and substantive interaction is the line the Department of Education draws between distance education and a correspondence course. Under 34 CFR 600.2 an online course counts as distance education only when a qualified instructor engages students in teaching, learning and assessment through at least two of five activities — direct instruction, feedback on coursework, answers to questions about content, facilitated discussion, or an activity the accreditor approves — on a predictable, scheduled basis, and monitors who falls behind. An institution where half its courses or students are in correspondence study loses Title IV eligibility.
Today an online learning office meets those requirements with a checklist and a term-end spot check of a handful of courses. It counts posts, and a count cannot tell an auto-released announcement or a “Nice work!” on every submission from teaching. A gap is found when it can no longer be fixed for those students, or years later at an audit. Regular and Substantive Interaction Monitoring reads every interaction in every online section each night, measures it against the course's own posted schedule, and puts a private, supportive note in front of the compliance director when a gap opens.
Estimated split for one section of about 360 instructor interactions a term (the demo college's volume), checked by hand and with the agents; the agents' time is not counted — our arithmetic, kept modest.
Seven agents read, label, re-check and measure every instructor interaction; the compliance director and the dean decide.
Brandvale Community College, Fall 2026, week 8 of 16. ACCT 101-03, Principles of Accounting I, is taught online by an adjunct, Paul Whitcombe, to 34 students, 22 of them with Title IV aid. Here is one week of RSI monitoring, screen by screen, in the working solution.
The Interaction reader took in 9,412 new announcements, discussion replies, feedback comments, inbox messages and live sessions, each matched to its section and instructor of record. The Classifier labeled 72% substantive. On the interaction pulse, one row per online section, solid beats teach, hollow beats are automated or empty, and every stretch over 7 days without a solid beat is shaded with its length.
41 of 1,148 online sections have gone more than 7 days without a substantive interaction. ACCT 101-03 is first, at 16 days.
Teresa opens the first hollow beat after September 26. It is an announcement released by the learning management system on its own, created on August 20 as a delayed post and attaching a module video recorded in Spring 2025. The label reads “Not substantive · automated or scheduled”, with its confidence and a link to the original.
“Module 6 is now open. Please watch the walk-through video and complete Homework 6 by Sunday at 11:59 p.m.”
The section record shows the last substantive interaction: a discussion reply on Saturday, September 26, at 9:14 p.m., explaining accrual versus cash basis. Since then there have been 46 instructor messages and none substantive: 21 automated announcements, 19 grade comments such as “Nice work!” and 6 logistics replies. The syllabus promises feedback on every homework set within 7 days, discussion twice a week and a weekly video walk-through; 4 students missed Homework 5 and 6 with no outreach found in the gradebook.
A feedback comment in PSYC 101-11 would meet the second element, feedback on coursework, but it was written by an embedded tutor, not the instructor of record, so the Classifier stopped at 0.58 and showed its steps. Teresa keeps the label, or counts it because the tutor is a qualified co-instructor on file, and records a reason code either way.
On a different model, the Skeptic labeled a random 5% of week 7 and agreed on 1,295 of 1,310 (98.9%), and agreed on 385 of the 412 posts near the line (93.4%). The 42 disagreements wait for Teresa, side by side, one at a time or in bulk with one reason code. Week 6's disagreements are why ACCT 101-03 shows now: Classifier v3.1 had counted auto-released recordings as direct instruction, and v3.2, approved on October 9 by Teresa and the learning systems administrator, re-read weeks 1 to 7 and changed 1,904 labels.
The Nudge agent's draft starts with Paul's own good work — his Homework 2 feedback showed a student where a $340 difference came from — then says what the record shows, offers three small ideas and help, and passed all six of the Skeptic's tone checks. Teresa edits one line and sends it at 8:12. She sends the other 33 clean drafts at 8:19; the two that needed an edit go at 8:26.
“This note is private. It goes only to you, is not shared with your chair, and is never used in evaluation.”
Paul's view shows his section's pulse, the note and what counts, taken from his own course: his Homework 2 feedback and his September 26 discussion reply count; the Module 6 announcement and “Nice work!” on 19 submissions do not. He can reply, book 20 minutes with the online teaching team, or ask for a second look at any label.
Paul posted on adjusting entries on Tuesday at 7:42 p.m., wrote feedback with content on 31 Homework 6 submissions on Wednesday, and held his live office hour. Across the college, gaps over 7 days fell from 41 to 15, and 27 of the 36 nudged sections closed their gap. Three sections remain on Title IV watch, with 63 students who have aid.
ACCT 101-03's record now shows 38 substantive interactions in week 8, elements i to iv met, the live office hour recorded (52 minutes, 11 attended) and all 4 students who fell behind contacted. Teresa prepares the section record: 412 interactions, 37 pages, student names replaced with codes and the “not for evaluation” notice on.
The dean releases the record for the program review file at 4:40 p.m.; the person who prepared it cannot. The release, Teresa's nudges and the Classifier's labels of Paul's 31 feedback comments sit on the same audit trail, with who, what, when, why and the evidence.
The same week, seen by the people who carry it — what their day looked like, and what it looks like now.
Seven agents read, label, re-check and measure every instructor interaction; the compliance director and the dean decide.
Takes in announcements (with the scheduled-post flag and attachments), discussion posts and replies, feedback and rubric comments, inbox messages and live sessions, and matches each to its section and instructor of record; posts by people on no roster go to the director.
Labels each interaction against the five federal elements of substantive interaction, with its confidence and the passage that decided it.
On a different model, re-labels a random 5% and every post below 0.80, blind to the first label, and checks every nudge for tone.
Reads each syllabus's posted schedule, measures gaps against the stricter of it and the 7-day default, and checks outreach to students who fall behind.
Drafts a private, supportive note: the instructor's own good work first, what the record shows with citations, two or three small ideas, an offer of help and the faculty agreement notice.
Records distance or correspondence status per section and term, watches the institution's 50% limits nightly, and tells financial aid what changed.
Keeps each section's dated weekly record, builds the export for each reviewer, replaces student names and IDs with codes, and seals and logs each export.
Send nudges, escalate, sign status changes. The agents propose; a named person decides.
The assistant answers with the sources cited and opens the screen it is talking about.
Why is ACCT 101-03 at the top of the pulse?
Its last substantive interaction was a discussion reply on Sat Sep 26 at 9:14 p.m. Since then the instructor has posted 46 times: 21 automated announcements, some with a Spring 2025 recording, 19 grade comments such as “Nice work!” and 6 logistics replies. None meets an element of the federal definition, and the syllabus promises feedback within 7 days. That is a 16-day gap. 22 of its 34 students have Title IV aid.
Which sections with Title IV students are on watch?
Two sections are past 14 days: ACCT 101-03 and HIST 201-02, with 41 students who have aid. Each stays “distance education” until you and the dean agree otherwise, and the college is far from the 50% limit.
Show me the nudges that need an edit before they go
34 drafts passed the tone check. Two did not: COMM 110-03 mentions a “department review”, which reads as a warning, and ECE 140-01 asked in August for messages by email, so I changed the channel. Every claim in a draft is cited to the record.
Prepare the program review record for ACCT 101-03
The record holds every interaction this term with its label and citation, the posted schedule, the gap and how it closed, the re-check results and every decision. Student names are replaced with codes and the “not for evaluation” notice is on. You prepare it; the dean releases it.
13 screens from the working solution, on its sample data. Pick one to see it large.
The thresholds every section is measured against are settings, versioned, and changed only when a second administrator approves.
| Setting | Default | Choose from |
|---|---|---|
| Gap that shades the pulse | 7 days | The posted schedule is used instead when it is stricter |
| Title IV watch | 14 days | Sections past it go on the financial aid watch list |
| Near the line | Below 0.80 confidence | Goes to the Skeptic and to a person |
| Independent re-check sample | 5% of last week's interactions | Plus every post near the line |
| Nudge drafts | Mondays, 8:00 a.m. | No instructor gets more than one a week |
| At least two of the five elements | On | Set by regulation (34 CFR 600.2), locked |
| Escalation | Only after a nudge, 7 days and the dean's approval | Faculty agreement, locked |
| Keep records | 5 years after the term | Legal hold; Title IV requires 3 years after the award year (34 CFR 668.24) |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
Sources: 34 CFR 600.2 (distance education, correspondence course, substantive and regular interaction, instructor) and the final rule, 85 FR 54742 (September 2, 2020; effective July 1, 2021) · 34 CFR 600.7(a)(1) and HEA §102(a)(3) (the 50% correspondence limits) · 34 CFR 668.24 (record retention) · U.S. Department of Education Office of Inspector General audit of an online university (September 2017), recommending the return of $712.7 million, as summarized by WCET · "demo" = seen in the working solution, on its sample data · "target" = design goal, measured in the live solution · "typical" = published figures and regulation · "estimated" = our arithmetic. Brandvale Community College, its faculty agreement and the people named on this page are fictional — characters and sample data in the working solution.
Regular and Substantive Interaction Monitoring is software for online learning offices that checks every online course for regular and substantive interaction between instructors and students. Seven agents read every instructor post, reply, feedback comment, message and live session each night, label each one against the five federal elements with the deciding passage quoted, and measure gaps against the course's own posted schedule. When a gap passes seven days, a private, supportive note is drafted for the compliance director to edit and send.
It is the Department of Education's test, in 34 CFR 600.2, that separates distance education from a correspondence course. Interaction is substantive when an instructor engages students in teaching, learning and assessment through at least two of five activities: direct instruction, feedback on coursework, answers to questions about content, facilitated discussion, or another activity the accreditor approves. It is regular when it is predictable, scheduled and monitored, including outreach to students who fall behind.
In the working solution, an instructor's feedback that shows a student exactly where a $340 difference came from counts, and so does a discussion reply explaining accrual versus cash basis. An announcement released automatically with a recording from an earlier term does not, nor does “Nice work!” on 19 submissions. The instructor sees these examples drawn from his own course.
No. The agents read posts rather than count them: automated and scheduled posts, generic praise, logistics and posts by tutors or publishers are each labeled as such, with the passage quoted. That is what a checklist or an activity count misses — a course can look busy with 46 messages and still have none that is substantive.
To federal student aid it becomes a correspondence course, and an institution where half its courses or students are in correspondence study loses Title IV eligibility. Here the Title IV agent records a section as correspondence only after the compliance director proposes it and the dean agrees, and then tells financial aid which aided students to review.
No. Under the faculty agreement settings, which are locked, interaction data is never a basis of evaluation or discipline, notes go only to the instructor and never to a chair, and every export carries a “not for evaluation” notice. Instructors see their own record and can ask for a second look at any label.
Yes. The agents read, label, measure and draft; they never send a note, escalate, change a status or release a record. Labels below 0.80 confidence go to a person, the compliance director sends every note, the dean decides escalations, a status change takes two signatures, and a record is released by someone other than the person who prepared it.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your syllabi, your faculty agreement and a term of your learning management system data — and it goes live once every quality gate has passed. We will walk you through it on your own sections first.
Every course file fixed before April 2027, and heard right by a screen-reader test.
1-min demoAcademic Integrity Case ManagementReferrals screened on how the work was written, with both sides in one packet.
1-min demoLearning Outcomes AssessmentAssessment reports built from scored student work, where every number opens its papers.
1-min demoWe'll run Regular and Substantive Interaction Monitoring on a term of your own online sections.