Nuix unveils major AI advancements in its Neo platform

Nuix has announced substantial advancements in artificial intelligence (AI) integration within its Nuix Neo platform, aimed at enhancing search and processing performance. The company disclosed these updates at its recent customer event, XLR8/24.

Nuix Neo, launched last year, is designed to assist users in swiftly and efficiently processing, identifying, and understanding complex data. The platform incorporates Nuix’s proprietary Cognitive AI, featuring over 500 pre-built models that facilitate high-accuracy data classification, categorisation, extraction, and risk scoring.

According to Jonathan Rubinsztein, Group CEO of Nuix, ownership of their language models ensures that they are thoroughly understood with explainable and defensible results. "We have chosen to invest in our own language models meaning that we know the dataset, understand the training, we know who trained it, and understand the design," said Rubinsztein. "We also have governance in place to update ethical controls as technology, regulations and societal expectations evolve. Meaning we can vouch for the responsibility of our AI."

The integration of these AI models has reportedly yielded significant improvements in search accuracy. Customers have observed up to a 600% reduction in false positives and up to a 20% reduction in false negatives. Tasks that previously required six weeks are now completed in just 12 hours on Nuix Neo.

In addition to these improvements, Nuix announced the forthcoming inclusion of an embedded deep learning framework within the Nuix Neo engine. This framework will enable the platform to interpret and utilise data as mathematical vectors, facilitating advanced functionalities such as semantic search. Unlike traditional keyword matching, semantic search allows users to search based on meaning, intent, and context. Furthermore, the deep learning framework will support the integration of public large language models (LLMs), enabling semantic searches across various languages and image formats, in addition to text.

"This is a game-changer for Nuix, and we are excited to announce a capability that very few offer," remarked Rubinsztein. "This accelerates enhanced search capability for our customers, enabling them to get to the truth in their data faster. And this is only just the beginning. Further development of Nuix’s deep learning framework will be directed at supporting the integration of other publicly available LLMs to drive future functionality in things like multi-lingual transcription and translation, retrieval augmented generation (RAG), text summarisation and advanced interactive user experience with AI chatbots."

Nuix Neo also supports the integration of customers’ own AI models, allowing them to interact with and enhance the platform. Results from these third-party models can be easily distinguished by colour coding and metadata, offering transparent differentiation between Nuix AI outcomes and others.

Additionally, Nuix has announced advancements in its Nuix Workstation product. The upcoming release, version 10, will deliver enhanced functionality, particularly in terms of engine support for additional data types.