Data Squared & Neo4j partner to deliver traceable AI systems

Data Squared has announced a strategic partnership with graph database company Neo4j to provide artificial intelligence solutions that address the challenge of AI "hallucinations" in both government and private sector operations.

The partnership centres on the integration of Data Squared's reView platform with Neo4j's graph database technology to build AI systems that prioritise verifiability and transparency. The companies are targeting scenarios where traditional AI models, particularly retrieval-augmented generation (RAG), can deliver convincing yet inaccurate responses, which undermine trust and inhibit decision-making.

Graph-based approach

Graph databases model information as nodes and relationships between entities - such as people, places, and events - instead of tables of rows and columns. This structural approach allows for the identification of patterns within large, heterogeneous data sets and supports the ability to answer complex questions that standard databases cannot always address efficiently.

Data Squared has developed a patented method anchored in this approach. Its system relies on Neo4j's ability to model and navigate complex relationships within datasets, forming the basis for reView's explainable AI capabilities. According to the U.S. Patent and Trademark Office, Data Squared's methodology is recognised as a new way to create "hallucination-resistant, explainable AI systems."

Traceability and explainability

By leveraging graph-connected evidence networks, every AI-generated answer on the reView platform is anchored to verifiable sources. This ensures that end users can trace how an AI system arrived at its conclusion or recommendation, meeting a requirement that is increasingly essential for government teams and critical applications in the private sector.

Jon Brewton, Chief Executive Officer of Data Squared, commented on the partnership:

Our partnership with Neo4j brings together the best of AI and graph database technology to deliver hallucination-resistant AI that can be verified, trusted, and understood. With Neo4j's graph backbone and our patented hallucination-resistant approach, we're helping mission-driven teams cut through the noise and take action with confidence. Decision-makers need to rely on AI that provides robust information security, not opaque black-box systems.

John Bender, Regional Vice President, Federal Sales at Neo4j, underscored the challenges their collaboration aims to address:

Data2 and Neo4j are solving one of the most pressing challenges facing organizations today: incorporating AI into their workflow with superior results and greater ROI. Our graph database and analytics technology helps eliminate AI hallucinations, while reducing resource consumption and increasing scalability, making it the ideal solution for organizations looking to AI to support complex, data-driven decision-making. We're excited to deepen our work together, empowering teams in the public and private sectors to transform data into knowledge and unlock insights and possibilities that weren't possible before.

Key capabilities and use cases

The reView platform delivers several capabilities as a result of this partnership, including patented hallucination resistance where every output can be traced to source material, visual traceability with complete source attribution, and unified integration of structured and unstructured data into harmonised graph models. The system's architecture supports enterprise-grade security with zero-trust and cloud-agnostic deployment and offers a natural language interface for non-technical users to engage with complex graph insights.

In practical applications, reView has been deployed to reduce analysis time in intelligence operations, provide traceable and explainable AI for federal agencies, improve asset management decision-making in the energy sector, and enhance fraud detection in financial services. In these scenarios, the platform claims to achieve near 99% accuracy in AI-generated insights, as compared to the 60-80% accuracy rates common to traditional RAG approaches.

The companies report that public and private sector organisations requiring full audit trails, regulatory compliance, and high-reliability AI decision support can access the platform through both Data Squared and Neo4j sales channels. Additional customer support is provided via fast-track implementation, technical integration, and proof-of-concept services using client data.