Building an AI Assistant for BI: The Good, the Bad, and the Ugly (repeat)
OVERVIEW
EXPERIENCE | In Person |
---|---|
TYPE | Lightning Talk |
TRACK | Data Warehousing - Analytics and BI |
TECHNOLOGIES | AI/Machine Learning, GenAI/LLMs, SQL Analytics / BI / Visualizations |
SKILL LEVEL | Beginner |
DURATION | 20 min |
This session is repeated.
How will generative AI impact Business Intelligence workflows? Software engineers who write code are rapidly adopting LLM-based chatbots. However, as I will explain in the talk, using this same kind of interface for data exploration and dashboarding comes with a unique set of challenges. This story will be told through our experience at Hashboard building an AI-powered assistant for data visualization. I’ll contrast the initial expectations of these features with the real-world use cases we saw from users. It turns out there is a constellation of tricky problems that need to be addressed for a Chatbot-like interface to succeed, including capturing business context outside of the data and guiding users to the appropriate datasets. These lessons inform a roadmap for how our BI tools should evolve to better leverage AI and ultimately make data insights more accessible to a broader audience.
SESSION SPEAKERS
Dan Eisenberg
/VP of Technology
Hashboard