Why this exists — and how the model differs from “every app owns a database.”
Want to run it? Install is on the home page.
App A App B App C App D
| | | |
v v v v
Notes DB Health DB Msgs DB Finance DB
(silo) (silo) (silo) (silo)Every application rebuilds accounts, storage, indexing, permissions, and sync. Each stores its own copy of your information. No single system understands the full picture.
Your data is fragmented, duplicated, and outside your control.
App A App B App C App D App E
\ | | | /
\ | | | /
+----------------------------------------+
| Your LastDB Database |
| encrypted / local / permanent |
+----------------------------------------+Applications become clients of the user’s database rather than owners of the data. That inversion is the whole product thesis.
Applications Notes · Health · Finance · AI · Email | v Shared Structures public interfaces, standardized | v Transforms local computation, deterministic | v Vector Embeddings semantic index across all data | v Encrypted Storage user-controlled, append-only
Computation runs locally. Raw data never leaves your control by default.
NEVER TRUST THE CLOUD
Data stays end-to-end encrypted. Cloud may store or transport bytes; it should not be able to read them.
REVEAL THE MINIMUM
Apps receive only what they need. Results live under access policies — nothing visible without permission.
SELF-MAINTAINING
You should not babysit schemas, migrations, indexing, or cleanup. The system organizes itself.
Structures are public and standardized: how data is organized, what you can query, how derived results are produced. Apps can work against any user’s database without custom migrations.
Deterministic functions attached to structures. Outputs are written back into structures — so they inherit the same access rules as anything else.
Structure: Messages Fields id · sender · recipient · timestamp · body Transforms inbox(user_id) conversation(user_a, user_b) semantic_search(query) unread_count(user_id)
One semantic index across notes, messages, code, and structured records. Search is meaning-based, not stuck to one app’s schema.
Systems can ask whether relevant information exists without exposing raw content or its origin. Collaboration without surrendering the silo model’s only remaining virtue: not dumping everything into a public feed.
LastDB is experimental and in active development. Early releases focus on local encrypted storage, apps on top of that node, semantic search, and the agent workflow we use to build it.