diff --git a/memory/2026-02-25-exploring-character-level-gpt-training-on-memory-d.md b/memory/2026-02-25-exploring-character-level-gpt-training-on-memory-d.md
new file mode 100644
index 000000000..6f33137ec
--- /dev/null
+++ b/memory/2026-02-25-exploring-character-level-gpt-training-on-memory-d.md
@@ -0,0 +1,11 @@
+# 2026-02-25 Session Notes
+
+## Exploring Character-Level GPT Training on Memory Database
+
+Nicholai proposed training the microgpt character-level model (Rust implementation at `references/microgpt/src/main.rs`) on personal memory data from `~/.agents/memory/memories.db` to see what the model "hallucinates" - essentially a funhouse-mirror experiment on his own agent memories.
+
+Investigated the memories database schema and found 5348 memory records. The `memories` table contains rich metadata: content (TEXT), type (fact/preference/decision/etc), tags, importance (0.0-1.0), confidence, category, timestamps, source tracking, and extraction status. The database includes FTS5 full-text search indexes and vector embeddings infrastructure.
+
+The character-level model is tiny (48 embedding dimensions, 2 layers, 4 attention heads) and expects line-delimited text input. The feasibility depends on extracting memories in a format the model can meaningfully learn from—raw content lines or structured summaries.
+
+No implementation work began; this was pure exploration of the concept and data shape.
\ No newline at end of file
diff --git a/memory/memories.db-wal b/memory/memories.db-wal
index 8a6c3dfb4..869e352fe 100644
Binary files a/memory/memories.db-wal and b/memory/memories.db-wal differ