← All scenarios

Partner and platform APIs

Every partner integrates against something different

Each partner integrating with the platform re-derives the same entities from a different undocumented endpoint.

Integration-heavy platforms

These are illustrative scenarios showing how the products are used, not customer case studies.

The same sequence, twice

The same sequence today, and with SynaptaGrid.

Today

  1. Partner needs data
  2. Bespoke endpoint added
  3. Docs written by hand
  4. Field change breaks partners

With SynaptaGrid

  1. Record type published
  2. Description generated from it
  3. Partner key scoped and limited
  4. Old version stays live

The situation

The contract is whatever the last integration needed. Documentation is written by hand and is stale by the next release, one credential is shared across partners because scoping it was never built, and taking a field away means calling everyone first.

How it is approached

Records live in one place, each record type and version with its own REST URL space, described by an API description the service publishes plus a schema and presentation description per published version. Partners get their own keys, scoped to particular record types, with their own rate limits, rotated or revoked without touching anyone else, and key holders are always served the published version rather than a working draft. Changes are pushed out as signed events with retries, a delivery log, and dead-lettering that can be redriven.

What changes

A new partner integrates against a description generated from the model rather than one written about it. Removing a field becomes a versioned change, with the previous version still active while partners move off it.