Data Normalisation Agent
Standardizes company names, contact fields, domains, and identifiers across all data sources.
What it is
"Accenture," "Accenture PLC," and "accenture.com" are the same company but a naive database treats them as three records. This agent resolves fragmentation issues at scale by standardizing names, domains, phone formats, and geographic identifiers across every data source.
How it works
Each record passes through a multi-step normalization pipeline. Company names are resolved against a canonical entity graph using fuzzy matching and domain anchoring. Contact names are parsed into structured components. Phone numbers are normalized to E.164 format. Geographic data is standardized to ISO country and state codes. Industry classifications are mapped to a canonical taxonomy. Each normalized record is written with a confidence score and a list of transformations applied.
How it fits into the pipeline
Runs after the Sheet Formatter Agent and before the Deduplication Agent. Without normalization, deduplication produces false negatives. Normalized records also improve Firmographic Enricher match rates. Output feeds the CRM Update Agent, Intent Signal Aggregator, and scoring model.