Squirro is a versatile generative AI tool designed for enterprise search, insights, and automation. It is built with an advanced AI framework, Retrieval Augmented Generation (RAG), that enhances the accuracy of the responses generated by larger language models (LLMs). This enhancement is made possible by incorporating external sources of knowledge to support the model's internal understanding. The SquirroGPT component of the tool uses Semantic Search to query the LLM, facilitating a more efficient and informed data retrieval process. When a user enters a prompt, SquirroGPT searches the knowledge base, including the ingested data and documents. Relevant information is then sent to the LLM, and the response is verified against the knowledge base before being relayed to the user. This ensures each answer is supported with evidence, and the refining process reduces inaccurate responses. The tool also specializes in providing accessibility to complex organisational data, with the capacity to define data sources and permission rights to suit individual business units. It enables users to interact with data without needing to open documents, providing more precise results by analyzing relevant paragraphs instead of entire documents. This tool offers enterprise-grade security and can be embedded for wide audience accessibility.
