Skip to content
D'Césares

Essay · 8 min read

Learning platforms with institutional memory

What changes when a course remembers every cohort that took it — and why that memory belongs to the teachers, not the tool.

ID

Isaac D'Césares

Every cohort leaves traces: the questions that came up in week two, the example that finally made tokenization click, the assignment nobody finished. Most platforms throw those traces away at the end of term.

What a course could remember

A course with memory keeps three things: the questions learners actually asked, the examples that worked, and the drop-off points where people quietly left.

  • Questions become the next cohort's FAQ — written by peers, not by us.
  • Examples are versioned with the material they explain.
  • Drop-off points are reviewed by facilitators before the next run.
Learning happens best when people build things that matter to them.

Where the memory lives

The simplest implementation is a plain file next to the syllabus. Inline code like cohort.questions stays readable; longer samples go in a block:

// Keep the memory with the material
const memory = await loadCohort("ai-literacy-1", { since: "2025-02" });
memory.questions.filter((q) => q.upvotes > 3);

None of this needs a new platform. It needs a decision about who owns what the course learns.

Try it with your next cohort

Start with one shared document of questions. That's the whole first version.