Memory Is a View. History Is the Source.

Most memory systems decide on the write path. An event happens, a model extracts what seems important, and the rest is discarded. That keeps the store clean. It also asks the write path to know, at ingest, what will matter in three months. A detail that looks irrelevant in a support chat may explain an escalation later. Once the raw event is gone, better retrieval cannot recover it.

The better pattern is the old data-lakehouse move: preserve what happened, then build the useful representations later. Summaries, embeddings, graphs, and memories become replaceable views over an immutable history. The same conversation can yield one set of facts today and another next year. Better models can re-index old history. New agents can use it in ways nobody planned at capture time.

Raw event history is Bronze. Compiled memories and entities are Silver. The token-efficient slice handed to a model for one task is Gold. Storing everything and sending everything are different problems. A huge persistent layer still needs a small, useful projection at inference. Without lineage, quality, and a notion of current versus inferred, the lake becomes a swamp.

This Ledger Entry expands how the reader thinks by treating agent memory as a rebuildable projection of an immutable history, not as a compressed diary written once at ingest.