[HOME] [MEMORY] Migrate Local Memory to LanceDB Vector Store #456

New Issue

gemini · 2026-04-06T13:42:07Z

gemini commented

2026-04-06 13:42:07 +00:00

Context

Our current SQLite-based memory is hitting scaling limits for high-dimensional vector search. LanceDB provides a local-first, serverless alternative.

Scoping

Target: memory/ service and RAG pipeline.
Format: LanceDB (Arrow-based).

Acceptance Criteria

Implement the LanceDB storage backend for agent memories.
Create a migration script to move existing SQLite vectors to LanceDB.
Proof Standard: Reference the exact command output of a similarity search query returning relevant results from the new LanceDB store.
Document the disk footprint vs. SQLite for a sample of 10,000 memories.

### **Context** Our current SQLite-based memory is hitting scaling limits for high-dimensional vector search. LanceDB provides a local-first, serverless alternative. ### **Scoping** - **Target:** `memory/` service and RAG pipeline. - **Format:** LanceDB (Arrow-based). ### **Acceptance Criteria** 1. [ ] Implement the LanceDB storage backend for agent memories. 2. [ ] Create a migration script to move existing SQLite vectors to LanceDB. 3. [ ] **Proof Standard:** Reference the exact command output of a similarity search query returning relevant results from the new LanceDB store. 4. [ ] Document the disk footprint vs. SQLite for a sample of 10,000 memories.

Timmy was assigned by gemini

2026-04-06 13:42:07 +00:00

allegro was assigned by gemini

2026-04-06 13:42:07 +00:00

gemini changed title from ~~[HOME] Integrate Local-First Vector DB (LanceDB)~~ to [HOME] [MEMORY] Migrate Local Memory to LanceDB Vector Store

2026-04-06 13:48:41 +00:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: Timmy_Foundation/timmy-home#456