Azure AIInstall

Summary

Fetches a previously submitted Azure Document Intelligence analysis result by model ID and result ID, and returns it as shaped JSON with extracted fields, confidence and pages.

Inputs

• Authentication (required) — API Key (default) or Entra ID (OAuth2). Selects which credential is used for the request.
• Model ID (required) — The model used when the analysis was submitted (e.g. prebuilt-invoice). Used in the request URL path. The node throws an error if empty.
• Result ID (required) — The result ID from the Analyze Document operation. Used in the request URL path. The node throws an error if empty.

Output shape

One JSON item per input item with documentType, confidence, fields (each field as {value, confidence}), pages, and _raw (the full analyzeResult).

Does a single GET to /documentintelligence/documentModels/{modelId}/analyzeResults/{resultId} (api-version 2024-11-30) and does not poll. Only the first document in the result is used for documentType, confidence and fields. documentType falls back to modelId when docType is missing. If the response has no analyzeResult (e.g. the analysis is still running or has failed), the output is just { _raw: }. Field value is taken from content, valueString, valueNumber, valueDate or value, in that order, so structured values (arrays, objects) may be missing from the simplified fields and are only in _raw.

Examples

Example 1: Retrieve an invoice analysis that was started earlier in the workflow or in another run.

Model ID: prebuilt-invoice; Result ID: expression like {{ $json.resultId }}; Authentication: API Key.

Example 2: Look up a custom model result and read its extracted fields.

Model ID: my-custom-model; Result ID: the stored ID; use the output's fields..value downstream.

Links

• n8n community nodes installation guide