Bot Confidence Level (AR Remittances)

Written by Uday Kumar Updated

Purpose

This article explains how Auditoria.ai calculates and applies document‑level confidence scores for AR Remittances.

The Bot Confidence Level helps you:

  • Assess the accuracy of remittance extraction

  • Flag remittances for manual review

  • Support data integrity in your AR cash‑application and posting workflows

1. Remittance Data Validation Process

For each Remittance document, Auditoria.ai performs three validation checks:

  1. Remittance Information Validation

  2. Amount Consistency Check

  3. Line Items Check (where automated line extraction is available)

The results of these checks are combined into a single document‑level confidence score.

1.a Remittance Information Validation

Fields checked (typical examples):

  • Customer

  • Entity

  • Payment Number / Payment Reference

  • Payment Date

  • Currency

  • Total Amount

  • (Optionally) Payment Type, Transaction Reference

Method:

  • Each header field’s extraction confidence is calculated and compared to internal thresholds (typically around 0.7).

  • Any critical field with confidence < 0.5 lowers the overall record confidence.

Effect:

  • If one or more critical header fields have Low confidence, the overall document confidence will not be High, even if other checks pass.

1.b Amount Consistency Check

Fields checked:

  • Total Amount (header)

  • Sum of line‑level applied amounts (where line items are extracted)

Method:

  • The system verifies that the header Total Amount is consistent with the sum of line‑level amounts, within allowed tolerances. 

  • If this check fails, the remittance is flagged for review.

  • Any amount validation failure sets overall document confidence to Low.

This check is purely mathematical and does not depend on individual field‑level confidence scores.

1.c Line Items Check (Automated Extraction Process)

When line‑level remittance details (for example, invoice references and applied amounts) are automatically extracted, an additional check is performed.

Method:

The system compares the sum of line amounts to the header Total Amount (or equivalent remittance total):

  • If sum(line amounts) > Header Total Amount → Confidence Score = 0.2 (Low)

  • If sum(line amounts) = Header Total Amount (within tolerance) → Confidence Score = 0.95 (High)

  • If line item duplicates were removed during processing, or sum(line amounts) < Header Total Amount → Confidence Score = 0.7 (Medium)

Additionally:

  • If line‑item confidence (for key line fields) is < 0.7, the remittance is flagged for review.

Effect:

  • Good reconciliation between line amounts and the header amount supports a High document‑level confidence.

  • Inconsistent or partial line‑level extraction results in Medium or Low document confidence, depending on severity.

2. Determining the Overall Document Confidence Rating

After the three checks above, Auditoria.ai calculates an aggregate confidence score and maps it to one of three document‑level ratings.

Table 1. Overall Document Confidence Rating

Confidence Level Criteria (AR Remittances) Visual Indicator Confidence Range
High No fields flagged; amount check passes; line‑item check score ≥ 0.7 Green ≥ 0.7
Medium Any remittance information flagged, or line‑item check score < 0.7; minor inconsistencies that still allow processing Yellow 0.3 < score < 0.7
Low Multiple issues, failed amount validation, or any critical field with confidence < 0.5; must be carefully reviewed by users Orange ≤ 0.3

The visual indicator appears in the “Bot Confidence Level” / “Confidence Level” column for each record within the AR Remittances Documents view and on the Document Information page.

3. Determining the Field‑Level Confidence Rating

In addition to the document‑level score, fields on the remittance can be visually highlighted based on their individual confidence scores. This helps users quickly identify areas of uncertainty.

Table 2. Field‑level confidence bands

Field‑Level Confidence Visual Indicator Confidence Range
High No highlight 0.7
Medium Yellow highlight 0.5 < score ≤ 0.7
Low Orange highlight ≤ 0.5

Color‑coding rules:

  • Color coding applies in both read and edit modes.

  • Only Medium or Low confidence fields are highlighted.

  • High confidence fields remain unhighlighted.

Typical fields that may be highlighted in AR Remittances include:

  • Customer

  • Entity

  • Payment Date

  • Currency

  • Total Amount

  • Payment Number / Transaction Reference

  • Line‑level applied amounts and invoice references

(The exact set of highlighted fields may vary. Contact your Auditoria.ai representative for the current list for your tenant.)

4. Flagging Fields in Remittances for Manual Review

When the bot detects Medium or Low confidence, fields are flagged for manual review.

  • Header level

    • Examples: Customer, Payment Date, Currency, Total Amount, Payment Number.

    • These fields are highlighted when their confidence is Medium or Low (see section 3).

  • Line‑item level

    • Examples: Invoice Number / Reference, Applied Amount, other line attributes.

    • These are highlighted when their confidence is Medium or Low.

Users should pay particular attention to:

  • Low confidence header fields that affect posting or reconciliation.

  • Low-confidence line items that affect how payments are applied to open invoices.

  • Any remittance where the overall Bot Confidence Level is Low, or where the amount consistency check failed.

For more information about Bot Confidence Level and how it applies in your environment, contact your Customer Success Manager or reach out to support@auditoria.ai.