Knowledge Graph Intelligence for Supply Chain Consultancy

Deployment

Tom Lorimer

Chief Executive Officer

The problem

‍

Can 15 years of scattered consulting work become a connected knowledge base that every consultant can query? Can it go one step further and have every answer cited and all confidential work kept secure? This was the challenge we set out to solve with an employee owned supply chain and logistics consultancy.

‍

They had built up a wealth of valuable knowledge, including pricing models, supplier intelligence and hard won delivery lessons. However, it was spread across hundreds of SharePoint sites, largely untagged and disconnected.

‍

01

Repeated work

New projects often started from scratch, because relevant past work was difficult to find.

02

Hard-to-access data

Much of the most valuable knowledge lived in spreadsheets, which language models can struggle to interpret accurately.

03

Confidentiality risks

Some client projects held strictly confidential information that could never be exposed to the wrong people.

04

Disconnected information

Keyword search could find a file, but couldn't connect a supplier, a project, a client and lessons learned.

‍

The Passion Labs Approach

‍

In four weeks, we designed a knowledge intelligence platform that runs entirely within the consultancy's own Azure tenant, building on its existing Neo4j, MCP and SharePoint sync prototypes.

‍

The result was a connected knowledge base, turning 15 years of scattered information into something consultants can actually query and use.

‍

‍

The platform combines knowledge graph and vector search results using reciprocal rank fusion, bringing together information based on both relationships and semantic relevance.

‍

Crucially, pricing and supplier figures are calculated in code rather than left to the language model, reducing the risk of inaccurate calculations. Every answer also cites its sources, so consultants can trace the information back to the original material.

‍

Confidential projects are excluded at ingestion, meaning permission safe answers are built into the architecture rather than relying on filtering after retrieval.

‍

‍

The impact

‍

15 years of institutional knowledge is now accessible to every consultant through a single intelligence platform. Past work becomes easier to find and reuse, answers are traceable to their sources, calculations are handled reliably and confidential client information remains protected.

‍

Now, 15 years of institutional knowledge sits in front of every consultant: cited, calculated and safe to use.

‍

< back to Projects

< previous

Next >