← Back to Case Studies

Big 4 Consulting Company

Transform tax and compliance advice at one point in time to continuously up-to-date guidance.

Challenge

Regulatory changes and judgements impact tax guidance issued for clients. Traditionally, this guidance is compiled by experts at one point in time. A first step towards “always-current” advice is to automatically surface relevant news from high-volume data sources. Can natural language understanding automate the identification of information that impacts client-specific issues?

Solution

Our system autonomously queries news feeds and compares new items against all existing guidance and tax memos. We built a custom knowledge graph derived from federal tax code, current treasury regulations, and other metadata — involving automated PDF text extraction of tens of thousands of documents and web crawling. We combined knowledge graph context with NLP-derived features (topic modeling) from news item text, and validated the hybrid similarity scoring model using ground truth provided by tax experts. Only relevant items alert the practitioner for review.

Results

The prototype demonstrated feasibility and led to a follow-on minimum viable product. The system acts as a highly accurate filter that understands both language and context, enabling the consultancy to move from “point-in-time” to continuously updated client guidance.