How the Spectare REST API flow works end to end

The Spectare API follows a three-step flow: qualify, assemble, and convert. POST to /api/qualify/{orgSlug} with a summary string describing the visitor context. The response returns a signed context token and an intent classification object containing type, confidence, audience, and stage. The qualify response no longer includes conversionUrlPatterns or defaultCurrency, as conversion goal resolution now happens entirely on the server.

With the token in hand, POST to the assemble route to receive a ranked set of content atoms and a variantId. The variantId is an opaque hash that identifies this specific assembly. Pass it back as excludeVariantId on subsequent assemble calls within the same session to ensure the visitor sees fresh content on each page without any atom IDs leaking to the client.

When a visitor converts, POST to /api/conversion with the signed context token, the URL where the conversion occurred, and the variantId from the assembly that was shown. Including the variantId closes the attribution loop so Spectare can credit the correct assembly in your analytics. You can also pass goalName, goalValue, and goalCurrency to link the event to a named goal and track revenue. All goal fields are optional and the endpoint accepts a simple page-reached conversion with no value if you omit them.

Conversion goal matching, currency resolution, and atom ID lookups all happen server-side. The browser never holds raw identifiers from your content database.