AI Auto Matching

Select a reconciliation run with unmatched items. The AI agent will examine your data, discover patterns, and automatically generate a multi-pass waterfall of matching rules — no manual rule configuration needed.

Powered by GPT-4.1
View Rule Builder View Suggested View Exceptions

How It Works

1
Data Sampling
Loads unmatched items from both sides of the reconciliation and identifies available fields
2
AI Pattern Analysis
GPT-4.1 analyses field values, amounts, dates, and references to find matching strategies
3
Rule Generation
Creates a waterfall pipeline: exact match → tolerance → fuzzy → aggregate → ML
4
Auto Matching
Applies generated rules, creates match pairs with confidence scores, and updates statuses

Reconciliation Runs

Select a run with unmatched items to analyse

Loading runs...

Select a Reconciliation Run

Choose a run from the left panel. The AI will examine the unmatched records, discover data patterns, and generate an optimal set of matching rules automatically.

Pattern Discovery
Identifies field relationships, value formats, and matching signals across both data sources
Rule Generation
Builds a waterfall of rules: exact → tolerance → fuzzy → aggregate, prioritised by confidence
Auto Matching
Applies rules to create match pairs, each with a confidence score and explanation