Skip to content
Scani docs

Why the three-tier deployment model

Scani supports three operational models with the same application schema.

  • Tier 1: operate the complete stack and manage upstream provider credentials.
  • Tier 2: operate UI, API, worker, database and S3; use one Scani Cloud key for platform processing and data-provider requests.
  • Tier 3: use Scani’s fully managed application.

Tier 2 separates durable data ownership from provider operations. The customer owns records, uploads and personal exchange/broker credentials. Scani Cloud manages AI and supported market/blockchain providers, processes the necessary inputs and returns results to the customer’s worker for local persistence.

Deployment selection is explicit through SCANI_DEPLOYMENT_TIER. Storage does not follow the presence of a cloud key. Both self-hosted tiers retain local S3; customer cloud keys cannot call internal storage or arbitrary-email endpoints. Authentication email uses constrained templates instead.

The provider capability interfaces keep application logic independent of local versus remote processing. The versioned processing.v1 contract sends minimal asset identifiers, public wallet addresses and requested document content. It preserves decimal strings, timestamps and incomplete-history warnings.

Tier 2 involves processing data outside the customer’s network. It offers control over durable storage, not an offline or zero-egress deployment.

See the tier model and Tier 2 overview for current requirements.