The latest articles on DEV Community by John Samuel (@codegallantx). https://dev.to/codegallantx https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F1165973%2F90a61e63-7561-4cce-861f-0da571ddc8a3.png https://dev.to/codegallantx en John Samuel Sun, 05 Jul 2026 16:56:52 +0000 https://dev.to/codegallantx/stateless-llms-are-a-problem-for-edtech-heres-how-i-fixed-it-with-a-graph-vector-memory-layer-1b1b https://dev.to/codegallantx/stateless-llms-are-a-problem-for-edtech-heres-how-i-fixed-it-with-a-graph-vector-memory-layer-1b1b <h2> The problem </h2> <p>Every AI study tool I'd used had the same flaw: total amnesia. Ask it about your exam today, come back tomorrow, and it's a blank slate. No memory of what you struggled with last week, what topics you already know cold, or when your next exam actually is.</p> <p>For "The Hangover Part AI" hackathon by WeMakeDevs, powered by Cognee, I decided to fix that for students specifically. The result is <strong>Summa AI</strong> — a learning companion built around persistent memory instead of stateless chat.</p> <h2> What it does </h2> <p>Summa AI is a Next.js frontend paired with a FastAPI backend, with Cognee as the memory layer running underneath everything:</p> <ul> <li>A chat workspace that keeps context across sessions instead of resetting every time</li> <li>Proactive exam check-ins pulled from stored exam data</li> <li>Knowledge gap detection — surfacing weak topics and missing prerequisites before they become a problem</li> <li>A Proficiency Hexagon in the dashboard, visualizing where you actually stand across topics</li> <li>Adaptive study plans that respond to what you already know, not a generic curriculum</li> <li>A <code>forget()</code> step for mastered topics, so the memory layer stays sharp instead of bloated</li> </ul> <h2> How Cognee powers it </h2> <p>I leaned on all four parts of Cognee's memory lifecycle:</p> <p><strong>remember()</strong> — every conversation, exam, and study artifact gets written into its own Cognee dataset. A <code>remember_conversation()</code> call stores the query + response summary; <code>remember_exam()</code> stores course, exam type, date, and topics; <code>remember_learning_progress()</code> stores topic-level scores.</p> <p><strong>recall()</strong> — this is the read path behind almost every feature: chat context lookup, exam reminders, saved artifacts, and the hexagon dimensions in the dashboard all route through it.</p> <p><strong>improve()</strong> — triggered right after new progress is recorded, or when a user gives feedback on a response, so the memory consolidates instead of just accumulating.</p> <p><strong>forget()</strong> — lets a student explicitly clear a mastered topic, or wipe a whole dataset, so old noise doesn't crowd out what's actually relevant.</p> <h2> What I'd build next </h2> <p>Deeper graph traversal for concept-map style prerequisite chains, and tighter feedback loops so <code>improve()</code> reacts to more signals than just explicit ratings.</p> <p>Built for students first — because the thing that should never have amnesia is the tool that's supposed to know how you learn.</p> <h1> WeMakeDevs #Cognee #TheHangoverPartAI #buildinpublic #nextjs #fastapi </h1> ai llm rag showdev