Doc Extraction helps PwC teams reason across complex spreadsheets

An internal capability that supports AI reasoning across large, multi-sheet spreadsheets used in audit, tax, and advisory work

Across PwC, spreadsheets remain a foundational tool for analysis, reporting, and decision-making. As workbooks grow in size and complexity—spanning multiple sheets, dense formulas, charts, and embedded files—they become harder to interpret consistently and efficiently. This complexity limits how effectively AI can be applied in spreadsheet-driven workflows.

Doc Extraction was created to address this challenge. It is an internal PwC capability that enables AI systems to navigate, interpret, and reason across large, enterprise-scale spreadsheets. The capability supports tasks such as tracing logic across sheets, surfacing relevant data, and explaining outcomes in context—helping teams focus their attention where it matters most.

Currently in use within PwC’s Assurance practice, Doc Extraction is being prepared for broader rollout across Tax and Advisory teams. The capability applies a multimodal reasoning approach that reflects how experienced practitioners work with spreadsheets—reviewing across tabs, integrating charts and supporting artifacts, and grounding outputs in verified data.

The experience is designed to fit into existing workflows. Teams provide spreadsheets and define the questions or areas of focus, and the system supports structured reasoning across the underlying data while maintaining traceability and governance.

By extending AI reasoning into one of the most widely used—and complex—data formats across the firm, Doc Extraction supports more consistent analysis and reduces the effort required to interpret large-scale spreadsheet models.