Agents as a user interface for data

“An intelligent data agent automatically converts user queries into SQL queries and returns clear answers. It speeds up information retrieval, enhances decision-making, and reduces the IT workload.”

Often, there’s a need to quickly retrieve information from a database without opening the correct report or a separate application view. I have long dreamed of an intelligent agent that allows users to interact with a relational database—just like talking to a colleague. My goal has been to build a functionality that understands natural language: the user’s question is automatically converted into an SQL query, which runs in the background against the database. The retrieved response is then returned to the user in a clear, natural format—without lines of code or digging through reports.

This setup might sound quite technical, but it isn’t. Imagine this scenario: a sales director wants to know which customer categories have generated the most revenue. They type the following question into the agent:

Which customer categories have generated the most profit?

They might receive an answer like this:

The customer categories that have generated the most profit are as follows:

• Novelty Shop: €123,220,010.30
• Supermarket: €14,360,495.80
• Gift Store: €11,683,614.10
• Computer Store: €11,560,346.80
• Corporate: €11,436,874.20

What if they wanted to see how this has changed over the years?

How has this changed over the years?

This brings us to the latest capability I have implemented in the agent—it responds like this:

Here is the visualization of how the profit generated by different customer categories has changed over the years.

If you need further analysis or details, please let me know!

The agent can also generate graphs for data visualization!

[If you are wondering about the years shown in the visualization, the demo uses the WideWorldImporters database, which contains order and sales records only from 2013-2016.]

Users can also choose to view the SQL query generated by the agent based on their natural language question.

By now, you are probably getting some ideas about how this kind of solution could benefit different user groups in your organization!

Faster and Easier Information Retrieval

Users can retrieve data without SQL knowledge—it’s enough to simply ask. The agent provides real-time answers, eliminating the need to wait for reports or increase the workload of data analysts. Manual report creation and data digging from different systems are significantly reduced.

Better Decision-Making

With up-to-date information readily available, decisions are based on facts rather than gut feelings. In the previous example, management can easily examine the profitability of customer segments or sales trends. Operational staff can also make daily decisions based on data without unnecessary delays.

Reduced IT Workload

Users no longer need IT or data analytics teams for simple queries. The IT department can focus on strategic development rather than constantly producing new reports. The need for static reports decreases as users can dynamically retrieve the information they need.

Improved User Experience

The agent enables natural language interaction without technical training. The user interface is intuitive and lowers the barrier to utilizing company data. Additionally, the agent can guide users with follow-up questions and suggest further analysis, making information retrieval even smoother.

These are just a few of the benefits—I’m sure you can think of many more!

If you’re interested in learning more about how this agent was implemented or would like to see a live demo, get in touch with us!

Article published originally in Finnish here.