Dataiku, known for its Everyday AI platform, has unveiled Dataiku Answers, a novel approach for data teams to construct Generative AI-driven chat systems using retrieval-augmented generation (RAG) on a large scale within enterprises. With Dataiku Answers, teams gain the capability to choose their preferred large language model (LLM), integrate their organization’s exclusive data through RAG methods, and create customized AI chatbots tailored for all departments across their enterprise.
Numerous companies have restricted access to public LLMs such as ChatGPT due to concerns regarding security and compliance, thereby hindering employees from leveraging the advantages of Generative AI in their daily operations. Even in cases where access is available, mainstream LLMs often lack the ability to access an organization’s proprietary data, resulting in unreliable insights and limiting the potential value of chat applications within the enterprise. Dataiku Answers addresses these obstacles by empowering data teams to effortlessly develop RAG-based chatbots fueled by proprietary content, facilitating accelerated insights into enterprise data and knowledge. The outcome is a multitude of secure, regulated, and scalable conversational AI chatbots that contextualize information and deliver pertinent and dependable responses to the organization’s distinct inquiries.
Sophie Dionnet, Global VP of Product and Business Solutions at Dataiku, emphasized, “Every organization can and should harness Generative AI to streamline operations and work more intelligently, and data leaders must be equipped to develop these applications with the appropriate level of transparency and reliability to mitigate risks effectively and at the right pace.” She added, “With Dataiku Answers, we merge the conversational capabilities of ChatGPT with the precision provided by RAG to provide data leaders with the enterprise-grade security, control, and visibility necessary for intelligent and responsible innovation.”
Dataiku Answers is built upon the Dataiku LLM Mesh framework, enabling data teams to connect with preferred LLM vendors such as Anthropic, AWS Bedrock, Azure, Databricks, Google Vertext, OpenAI, and others, as well as vector stores like Pinecone, to construct their AI chatbots. Alternatively, they can utilize self-hosted LLMs. Subsequently, they can easily construct RAG pipelines to grant the chatbot access to proprietary content, ensuring accurate and tailored responses to the organization’s needs. The Dataiku LLM Mesh comprises dedicated components for AI service routing, personally identifiable information (PII) screening, LLM response moderation, as well as performance and cost tracking, enabling trusted deployment of Generative AI.
Dionnet further remarked, “Data leaders have been inquiring about how they can streamline the deployment of RAG-powered chatbots across their organizations. Dataiku Answers already caters to this demand for over 20 global retailers, manufacturers, and other entities. In just a few weeks, we’ve witnessed them fulfilling a range of requirements, from implementing a company-wide LLM chat for corporate documents to developing domain-specific chatbots for investor relations, procurement, sales, and other teams.” She concluded, “Employees can pose questions as they would to ChatGPT and feel assured they are receiving reliable responses. Meanwhile, data teams gain comprehensive visibility and control over usage, costs, and quality. It’s a win-win situation for all.”
