> ## Documentation Index
> Fetch the complete documentation index at: https://docs.silnahealth.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Payor Inference Overview

Payor Inference extracts and confirms insurance plan details so you can create accurate patient plans in Silna. It supports
two workflows: processing insurance card images and confirming plan details directly with the payor.

Using Payor Inference before creating patient plans is optional, but recommended. It catches incorrect member numbers,
mismatched payors, and other data quality issues before they cause downstream failures in benefits checks and prior authorizations.

## Recommended Flow

1. Create a patient (`POST /v1/patients`)
2. Run Payor Inference — `BASE` if you have an insurance card, `CONFIRMATION` if you already have plan details
3. Poll until `COMPLETED` (`GET /v1/payor_inference/{id}`)
4. Create patient plans — review any `confirmation` results, but you can always proceed regardless of the outcome

## Strategies

The `strategy` field on the create request determines which processing steps run and what inputs are required.

### BASE

Use `BASE` when you have an insurance card. Silna will parse the document, identify the payor, confirm the extracted
details directly with the payor, and return the results along with any corrections. This is the most thorough strategy
and typically takes a few minutes.

**Required inputs:** `patient_document_file_ids`

### NEAR\_REALTIME

Use `NEAR_REALTIME` when you have an insurance card and need results quickly. Silna will parse the document and identify
the payor, but will not confirm with the payor. Results typically come back in seconds.

**Required inputs:** `patient_document_file_ids`

### CONFIRMATION

Use `CONFIRMATION` when you already have plan details and want to confirm them directly with the payor without processing
any documents. This is useful when plan information was entered manually, came from an EHR, or you want to verify
existing data before creating a patient plan.

**Required inputs:** `individuals` array with plan details to confirm. See
[Create Payor Inference](/api-reference/v1payorinferenceresource/create-payor-inference) for input details.

If confirmation is not available for the specified payor, the request completes immediately with `confirmation`
set to `null`. You can proceed with creating the patient plan — confirmation is not required.

## Understanding the Confirmation Response

When the `BASE` or `CONFIRMATION` strategy is used, each plan in the response may include a `confirmation` object
with the following fields:

* **`result`** — `CONFIRMED` or `NOT_CONFIRMED`
* **`requested`** — The values that were sent to the payor (payor, member number, patient demographics)
* **`detected`** — The payor's version of the details. Present only when `CONFIRMED`. If `detected` values differ
  from `requested`, review the differences — the payor's values are generally more reliable.
* **`error`** — Why the payor could not confirm the plan. Present only when `NOT_CONFIRMED`. Error types are `AAA`
  (payor rejection), `CAQH_CORE_ERROR` (clearinghouse error), or `INTERNAL_ERROR`.

### What to do with confirmation results

* **`null`** — No confirmation was performed. Proceed with creating the patient plan using the returned values as-is.
* **`CONFIRMED`** — You're good to go. Create the patient plan using the result values. If `detected` values differ
  from `requested` (e.g., a corrected member number or date of birth), verify the corrections look right before
  proceeding.
* **`NOT_CONFIRMED`** — This does not necessarily mean the plan details are wrong. The payor may have been temporarily
  unavailable, or there may have been an internal issue. The `error` field may provide a clue, but it's not guaranteed
  to be actionable. Double-check what was entered, make any corrections if needed, and retry — or go ahead and create
  the patient plan anyway. Confirmation is not required to proceed.

See the [GET endpoint](/api-reference/v1payorinferenceresource/get-payor-inference) for full response examples.