AI/BI Genie

Converse with your data

Intelligent analytics for business teams

AI/BI Genie is a conversational experience for business teams to engage with their data through natural language. Genie leverages generative AI tailored to your organization’s business terminology and data and continuously learns from user feedback. This allows you to ask questions the same way you would ask an experienced coworker and receive relevant and accurate answers directly from your enterprise data.

With Genie, users can self-serve and obtain answers to questions not addressed in their Dashboards without having to learn complicated tools or rely on expert practitioners. Backed by the lightning-fast performance and scale of Databricks SQL, users get answers immediately, all while upholding the governance and controls established in Unity Catalog.

Go beyond your dashboards

Genie is a no-code, self-service experience for your business teams to go beyond the answers in their Dashboards and ask any question of their enterprise data in natural language. Leveraging generative AI and large language models (LLMs), Genie allows users to have a conversation directly with their data without the need to learn BI or SQL tools. Users can ask questions and dive deep into their data to get instant answers and visual insights. Because Genie always remembers your conversation history and context, you can have a continuous dialogue, allowing you to get to the root of the problem.

Continuously improving AI from real-time user feedback

AI that learns from continuous feedback

At the core of Genie is a compound AI system that utilizes an ensemble of AI agents to reason about business questions and generate useful answers in return. This agentic reasoning enables Genie to refine its understanding of the questions you ask. If the AI is unsure how to answer, it will ask you for clarification or for an example to improve its knowledge. This iterative approach teaches the AI more about your data and business semantics to continuously improve the relevancy and accuracy of answers.

“Kythera Labs and Databricks AI/BI Genie simplify how healthcare business users find reliable answers with natural language and without coding.”

— Ryan Leurck, Chief Analytics Officer, Kythera Labs

Expert curation for trusted decision-making

For commonly asked questions and well-established business metrics, expert practitioners can curate instructions and certified answers for use with Genie ahead of time. Instructions can be added as example queries or snippets of plain text to help the AI answer business-specific questions with more precision. Certified answers are predefined recipes added in the form of parameterized SQL queries to reduce the likelihood of users receiving responses that are misleading or incorrect. When triggered, users will see a Certified Answer icon in the UI so they can have full confidence in the response generated.

“Genie is remarkable in its ability to understand the concepts and relationships that tie our operational and engagement data in ways that are unique to Accolade.”

— Kapil Ashar, VP of Enterprise Data and Clinical Platforms, Accolade

Activity monitoring to understand usage and adoption

Once deployed, users can monitor Genie activity to understand usage patterns. Every event is captured — including the questions asked, who asked them and when, the real-time status of any query, and how users rate the answers generated to help improve them over time. Genie monitoring gives you full visibility into every interaction to ensure adoption and overall success within your business.

Self-service insights with governance and security

With users across the entire organization directly accessing your data and AI assets, it is essential the insights they receive are governed and secure. Genie is tightly integrated with Unity Catalog to ensure any answer returned is secure and complies with the access policies established on the underlying data. This protects privacy, ensures regulatory compliance and strengthens security by keeping sensitive data hidden from view.