Forge - AI Agent
Normalize nested JSON, classify PII, and generate dbt models — all inside your warehouse
Publisher
Foxtrot Communications
Product Details
This agent normalizes nested JSON in a data warehouse without hundreds of hand-written UNNEST queries or brittle pipelines that break when schemas change. Pointed at a table, it discovers every field at every depth, infers types, classifies sensitive data, and generates clean dbt models that turn everything into queryable tables, while the data stays in the warehouse.
Discovery uses breadth-first traversal across all rows, including dynamic keys, and draws on training from more than 40,000 public API schemas. Generated dbt models include typing, relationships, and documentation. Each field is classified by Excalibur, a proprietary ML model that identifies PII, financial data, health information, and over 50 sensitive categories with 95% or better accuracy. Teams can apply hashing, masking, encryption, tokenization, or redaction at field level, and generate classification reports, audit logs, and lineage documentation for GDPR, CCPA, HIPAA, and SOC 2 work. The SQL runs natively in BigQuery, Snowflake, or Databricks.
Key Use Cases
In-Warehouse Nested JSON Normalization
Traverses complex nested JSON payloads directly inside cloud data warehouses to generate typed, production-grade dbt models without fragile manual UNNEST SQL.
Automated Data Discovery and PII Masking
Employs proprietary machine learning classification to discover sensitive data categories and apply field-level masking or tokenization for regulatory compliance.
Explore detailed deployment path
Requires Gemini. Access integration prerequisites, specialized agent configuration guides, and implementation documentation.