Generative AI (genAI) was supposed to quickly disrupt and radically change the way we analyze data using our existing BI tools. Maybe someday, but not yet. Unlike other genAI use cases (e.g., content generation, writing automation, etc), genAI is not disrupting the enterprise business intelligence (BI) software market. Instead, GenAI is adding valuable incremental capabilities, but it also necessitates building genAI guardrails at the enterprise level that brings into question the net value of the effort.
For now, BI infused with genAI capabilities can:
- Speed up SQL and other query and executable scripts development (CoPilot).
- Speed up building Q&A UI – BI chatbots – on data.
- Speed up building and enriching BI semantic layer (with data type semantics, synonyms, etc).
- Transform unstructured (text, images, audio) and semi-structured (documents, forms) to structured data.
However, this comes with several caveats:
- Most BI vendors we reviewed in our Next Stages Of Enterprise GenAI Infused BI Evolution research still map SQL generated from natural language questions to pre-built SQL that has been vetted and can function as a trusted source of data and insights.
- Investing time and effort in building and enriching your semantic layer becomes even more important – the richer the semantic layer, the more accurate the resulting SQL.
- Unless you are buying a platform or an application from a vendor, you’ll have to invest in genAI guardrails such as prompt engineering or RAG to assure an adequate level of accuracy.
Have more questions? Setup a call with me.
Source link
lol
