Stop Renting Your AI's Memory — Dylan Couzon, Qdrant
- Source
- youtube.com
- Author
- AI Engineer
- Date
Context
An assistant's stored memories can remain on the device that uses them. Qdrant Edge is a vector search engine that runs inside an application's own process with local storage and no separate server. Vector search finds stored items by similarity of meaning, so the application can retrieve relevant material without contacting a search server.
That capability underlies this talk's argument for owning an assistant's memory. The publisher describes memory as writing, retrieving and forgetting information. Qdrant's documentation says Edge is designed for environments with limited or intermittent connectivity and can optionally synchronize with a Qdrant server, so a system can be built around local storage without excluding later sharing. The documentation also describes Edge as in beta, so its interface may change.
- Couzon's framing: owning gives you autonomy, owning memory gives you continuity, and the memory that makes an assistant yours still usually lives in someone else's cloud.
- He maps the stack onto a computer, with the model as CPU, as RAM and memory as disk, and argues retrieval beats dumping a big folder of markdown into the prompt.
- The fully offline demo runs Qdrant Edge, embedded vector search, on a drone that builds its own searchable memory, with semantic queries under 1 ms in a 15 MB footprint.
- Memory can optionally sync to the cloud as a shared 'hive mind', which he presents as opt-in sharing with consent.
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
- inference — Running a trained model to get answers — the phase where AI is actually used, as opposed to trained.
- context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
videoEveryone Has the Same AI Models. Here's What Sets Agents Apart - Richmond Alake, OracleAI Engineer
videoNo Memory, No Harness: Why the Database Is the Last Line of Defense — Kay Malcolm, OracleAI Engineer
videoHow Inference Keeps Up with Our Insatiable Need for Tokens — Cerebras & Gimlet LabsAI Engineer
Checking sign-in…
Loading comments…



