Back to Documentation

Hub-Spoke Architecture

Learn the two-tier federated model powering LOB Brain.

LOB Brain introduces a federated model to compartmentalize local context while allowing teams to consolidate knowledge efficiently.

Spoke Nodes (Edge)

Local Engine: The Spoke node is a lightweight, native Rust binary running directly on each developer's machine.
Zero Cloud Data: Memory fragments tagged as secret or internal never leave the Spoke node. Developers maintain absolute custody over their local memory vaults.
SQLite/FTS5: Spoke data queries are processed locally at lightning speeds via SQLite's native FTS5 full-text search engine.

Master Hub (Central Core)

Federated Index: The Hub acts as a centralized repository aggregating public memories across the organization for cross-team reuse.
Explicit Sync: AI memories in a Spoke must be explicitly flagged with sensitivity: "public" to be synchronized with the Hub.
Collaborative Context: When a developer's agent encounters an unfamiliar project or domain, the Spoke fetches reference context from the Master Hub via a secure protocol.