A presentation by Graphwise and EY demonstrated how large organizations can transform their approach to knowledge management through strategic implementation of taxonomy management and knowledge graphs
Organizations with hundreds of thousands of employees face a fundamental challenge in knowledge management. Content flows in from multiple sources, creating data silos across different systems and applications. Without proper classification and tagging, valuable organizational knowledge becomes effectively invisible to those who need it most. Poor search results, duplicate content creation, and other inefficiencies compound quickly, representing significant costs in both time and missed opportunities.
At Knowledge Summit Dublin 2025, held at Trinity College Dublin on June 29th-30th, enterprise leaders gathered for an innovative conference designed to prioritize the exchange and creation of tacit knowledge over traditional presentations. The event’s theme “Humans In The Loop” explored building networks and skills to elevate knowledge and integrate AI with human connection at the core.
A presentation by Helmut Nagy from Graphwise and Arup Vidyerthy from EY provided valuable insights into tackling the challenges of knowledge management at scale. Their talk, “Connecting the Dots: Building a Collaborative Knowledge Hub with LLMs and Graphs,” showcased how a strategic blend of taxonomy management and knowledge graphs can create a scalable, AI-ready foundation for large enterprises.
A reimagined approach to knowledge management
In his presentation, Arup Vidyerthy explained that with almost 400,000 people, EY faces a significant challenge in managing the massive influx of content. According to Arup, the previous system created significant knowledge management bottlenecks. Content wasn’t properly classified or tagged, knowledge managers spent a lot of time on manual review processes, and employees struggled to find relevant expertise within meaningful timeframes.
Arup outlined EY’s “Discover Reimagined” initiative for digital transformation effort aimed at redefining how the organization shares, manages, and retrieves its content by introducing a new, AI-driven technology. This initiative focuses on four core objectives: findability, integration, performance, and content quality.
During the presentation, Arup described EY’s new strategy as combining best-in-class software and technology to create an ecosystem where everything is personalized. “Whether you come from a particular sector or service line, the content we present to you is personalized, but most importantly is integrated with the tools and texts that client-serving practitioners use daily.”
This approach is based on three pillars:
- Harvest: Knowledge is harvested through an intuitive submission experience that allows all EY employees to submit a diverse range of content.
- Review & Optimize: All required review functions including sanitization, auto-tagging, human review, optimizations, and approvals are implemented via a series of orchestrated workflows.
- Storage & Distribution: All approved assets are stored in a content management system. A knowledge platform portal allows all EY employees to search, access, and download highly curated knowledge assets.
Current capabilities and looking ahead
As Arup detailed EY’s current capabilities of their knowledge management platform, he explained that in terms of AI use, they already have intelligent content sanitization and personalization based on user profile and behavior data in the platform. They were still working on using GenAI to refine and optimize the taxonomy, on semantic recommendation, and on improving the auto-tagging accuracy by utilizing the knowledge graph.
The AI roadmap included hybrid RAG and GraphRAG use cases, deep integration of the knowledge graph with LLMs, and a RAG data processing framework for handling all approved knowledge assets at scale. When asked about performance improvements, Arup reported that when EY re-platformed the knowledge platform, they saw uptake in usage and adoption – quite often around 50 to 60% improvement.
In terms of taxonomies and ontologies, EY now has a centrally curated taxonomy, they have tagging against a reference taxonomy for our content, and they can do ontology-based searches. Arup also shared that their knowledge graph pilot has been completed and has moved into production implementation. What’s coming next is semantic and hybrid search with knowledge graphs, semantic recommendations, and GraphRAG with knowledge graphs.
Managing content at scale
The presentation highlighted the crucial role of the Graphwise Platform in this initiative. As Arup detailed, the platform is central to their operations, maintaining EY’s global taxonomies and automating the content tagging process.
“We have multiple taxonomies – some global taxonomies are maintained in the platform. Most importantly, when content comes in, we have a process orchestration component integrated with Graphwise PoolParty that looks at the reference taxonomy, scans our content, and tags it.”
Once content is approved in the content management system, all the finalized tags flow directly to Graphwise PoolParty to build a knowledge graph. This creates a rich representation that stores both the content metadata and extracted text while maintaining connections back to the original sources.
But the real power comes from the unified approach. Every application across the company uses the same centrally managed taxonomy to classify content. This consistency automatically builds the knowledge graph and breaks down data silos, enabling all GenAI applications to work from a shared, structured foundation.
A multidisciplinary AI approach
Helmut Nagy also shared Graphwise’s perspective on what made this work so successful. He noted that the knowledge management team at EY had already built a foundation of high-quality, richly-tagged content. Instead of building from scratch, this existing expertise was used as the foundation for the knowledge graph and AI systems.This multidisciplinary approach, combining EY’s deep knowledge expertise with the capabilities of the Graphwise Platform, created something more powerful and effective than either could achieve alone.
Helmut Nagy also shared Graphwise’s perspective on what made this work so successful. He noted that while AI should involve multiple disciplines, most companies struggle to achieve this integration. Often, AI teams simply want access to data without considering the quality work that knowledge management teams put into curating and structuring content.
To wrap it up
Arup and Helmut’s presentation demonstrated the realities of implementing knowledge systems across global operations. The transition from manual content processing to automated knowledge management shows how organizations can create intelligent content platforms.
The combination of EY’s high-quality content and Graphwise expertise provided the best support for both current knowledge management needs and future AI applications. This approach transforms knowledge management from a cost center into a strategic asset that enables better decision-making across the organization.