The students who disappear quietly are the ones a weekly view catches
When you teach a large number of students, you watch those who ask questions and those who clearly fail. You only become aware of those who quietly stop handing in their work when a deadline approaches.
A weekly overview solves that problem. It does not make evaluations. It gives a list of those who have missed something, so an educator can contact them. That contact is of some help.
Change is what matters. A student who has always submitted late is an established pattern. A student who submitted everything until three weeks ago is a different situation. Only the comparison reveals this.
The text contains no conclusions about a student. Missed assignments and attendance figures are factual. The reasons behind them cannot be inferred from the data. The result is a starting point for a conversation, not a judgment.
How the ActionFlow runs on the canvas
Each week the Start Scheduled node is triggered. The HTTP Request node then reads the submissions, attendance, and assessment completion from your learning system.
The Math node calculates the amount still outstanding for each student and the difference relative to that student's previous-week figures. Comparing each student to themselves rather than to the rest of the cohort is what makes the approach both humane and accurate.
A Filter node produces a list of students with outstanding items. That list is much shorter than the cohort, and it is the only part an educator needs.
The Generate Object node categorizes findings such as outstanding submissions, attendance falling below your threshold, and changes from an established pattern. It presents them as separate categories rather than as a single risk list.
The Generate Text node produces the educator's summary by indicating what is missing and what has changed, without trying to explain why.
The Slack or Email node sends it only to the educator. Nothing is sent to students as a result of this process, and nothing is entered into a student's record.
A second Start Scheduled flow provides program leads with a cohort-level view in which individual names are suppressed, so they can see course-level issues without naming any specific person.
Nodes this ActionFlow uses
- Start Scheduled: runs weekly ahead of teaching.
- HTTP Request (Util): reads submissions, attendance, and completion from the learning system.
- Math (Util): computes outstanding items and change against each student's own history.
- Filter (Data): reduces to the students with something outstanding.
- Generate Object (AI Core): returns outstanding, attendance, and change as separate categories.
- Generate Text (AI Core): writes the educator summary with no interpretation of cause.
- Slack (Communication): delivers to the educator only. Nothing reaches students or records.
Frequently asked questions
Start building AI workflows
Create a free account, open a template or a blank canvas, and run your first ActionFlow.
Newsletter
Get product updates
New nodes, agents, and product notes. We send mail only when we have something worth opening.