Aller au contenu principal

Hub & Spoke Architecture

The distributed intelligence model of Synap


Overview

Synap uses a Hub & Spoke architecture where:

  • Data Pod = Central Hub (User's private data kernel)
  • External Services = External Spokes (Intelligence providers, integrations)
  • Hub Protocol = The universal language connecting them

This model ensures Data Sovereignty while enabling Unlimited Extensibility.


The Hub (Data Pod)

The Data Pod is your personal server using the Synap Data OS. It is:

  • Private: You own the infrastructure (or rent a private instance).
  • Authoritative: It holds the "Golden Record" of your data.
  • The Gatekeeper: No data leaves or enters without permission (via Events).

The Spokes (Services)

Spokes are external systems that provide capabilities without owning data. They only access what is explicitly shared for a specific request.

Examples:

  • Intelligence Service: A GPU-heavy cluster running Llama-3 that summarizes your notes.
  • Calendar Bridge: An n8n workflow that syncs Google Calendar events.
  • Voice Interface: A mobile app that captures audio and sends it to the Hub.

Why This Architecture?

FeatureMonolith (Traditional)Hub & Spoke (Synap)
PrivacyData mixes with othersData isolated in Hub
AIGeneric, closed modelsPlug-in any model/service
ScaleHard to scale user-specific logicSpokes scale independently
Lock-inHighLow (Swap spokes anytime)

Communication: The Hub Protocol

The Hub and Spokes communicate via the Hub Protocol, a standardized, secure bidirectional protocol.

  1. Hub requests Expertise ("I need this note summarized")
  2. Spoke requests Data ("I need the note content")
  3. Hub grants Access (Temporary, read-only token)
  4. Spoke returns Insight ("Here is the summary")

This ensures the Hub always controls the data flow.