In recent years, Large Language Models (LLMs) have revolutionized the field of natural language processing (NLP) with their impressive ability to understand and generate human-like text. But did you know that these models can also be deployed for numerical data analysis?

Imagine you are the sales manager preparing for a meeting with a key account or you are the CEO that is about to go in a meeting with Private Equity on last month’s performance. Often you still have specific questions, but while the answers are in your own databases, they are often not available at your fingertips. You can call a colleague to find the answers, read the last reports or open one of many dashboards. Even if you have found the answer, it probably triggers another relevant question.

Open-source tools in combination with powerful LLM’s can now be deployed in organisations to let users have a chat with their business intelligence data. LLM’s do not only reply in natural language, but they can also deliver accurate SQL queries. It allows for a dynamic interaction with a database with instant results. Let the tool also visualize the answer in a useful graph that you can immediately use in presentations.

We see business intelligence chat agents not as a replacement of what is currently used, but as a next step in the evolution, a different way of interaction with your data. It’s not only about the technology. Knowing what questions to ask and how to train the model is equally important. This is called prompt engineering: the process of designing and refining input prompts to optimize the performance of language models. The quality and structure of the prompt can significantly influence the model's responses, making prompt engineering a vital aspect of working with LLMs.

Contact us if you want to know more about the role of AI in business intelligence, but also if you have other data challenges to solve first.