The Document Model

How this memory is actually stored in MongoDB

Why this matters: Storing JSON is one thing. Making the distributed document the fundamental, native unit of your entire architecture is another. Notice the embedding for vector search, stability for decay, metadata for flexible context -- all in one document, one collection.


        
🍇

Member Berry

Waking up the brain...

Memory Sources

Memories used to generate this response

How It Works

🍇
Member Berry

What if memory was a document?

An AI agent's memory isn't just text. It's embeddings, emotion scores, decay curves, graph relationships, and temporal context -- all different shapes of data. The document model handles every one of them natively, in a single collection. No vector sidecar. No graph database. No migrations.