Exhibitor News

In this article, let’s look at traditional workflows for building a BI solution (and some common bottlenecks along the way) and how AI is helping to change that and potentially improve our efficiency in delivering a quality solution to end users.

Traditional Workflows — How It's Usually Done

Here’s the daily reality for many of us:

Let's say a client wants a dashboard to view some sales performance data. First comes the meetings to gather their requirements and agree on a deliverable. Then comes the actual challenge of identifying and sorting through the noise of business data, seeing what is truly relevant and linking multiple sources of data together. Simultaneously, clients want to see what the result would look like, at the very beginning of the process, so we take meticulous care to create a mockup/wireframe of the dashboard.

Then comes the important feedback that may come through many rounds of meetings. 

“What if we change this to a line chart and view the trend over time?” 

“Let’s change these colours to match our new colour palette” 

“Spoke to the Sales team. They would actually like to see another visualisation showing the KPIs with Year-on-Year and Month-on-Month growth rates. Can we see what that would look like?”

After many reiterations and manual changes to the wireframe (which often can be time-consuming), the development of the dashboard actually begins. Post development and deployment of the dashboard, which too would have a couple of back and forths, finally comes to presenting the findings and insights to the client.

Usually, this process can take up anywhere from 2 weeks to even months and months of work in progress to deliver the end result. Whilst these BI workflows are robust and practised widely, there are a multitude of opportunities to improve the efficiency and effectiveness of our work through the use of AI.

Upskill Workflows — How It's Usually Done, But Better

In recent years, many different areas of work have adapted or even used AI to replace certain job roles. In the realm of BI workflows, whilst it’s quintessential to have BI Engineers to drive the whole process, the use of AI as a helping hand can enable certain tasks to be automated, freeing up valuable time to focus on development, deployment and maintenance of dashboards.

Dashboard Designing — Redesigned

The first time-consuming process that would benefit from AI is the whole design process. When dealing with clients, they would like to see what result would look like. And it’s common to have a change of mind, when it comes to the type of charts, colours used and layout, the addition and/or removal of visuals. Although accommodating all these requests in the past would have taken a significant number of days, now it can be done in a matter of minutes.

Using AI tools like Figma Make or Claude, your design conversation just got faster and better. Simply create a prompt describing your requirement, a rough outline of the type of charts anticipated and a colour code to be followed and generate a cohesive wireframe to satisfy your clients. From here onwards, you can get involved in the nitty-gritty details with your client, without having to spend long hours coming up with wireframe designs. AI can do the ‘hard work’ for you, whilst you focus on the ‘smart work’.

Tools like Claude can even be useful in later stages to analyse an existing dashboard design and layout. If a built dashboard looks too cluttered or if the colour combinations are just not working, you can generate an alternative ‘look and feel’ for your dashboard. Ultimately, it comes down to your expertise and sense for how the end product will actually look like, but utilising AI in this process helps to get there faster.

A Dashboard that Talks Back

Once all developments are done and the dashboard is published for the client to use, it’s usually a static output. The process usually ends there. Insights still have to be derived by manually analysing the charts and KPIs. But what if we can make it dynamic? What if your dashboard can talk to you and give you insights?

From the many upcoming and rising tools, we have the Databricks AI/BI and Genie function, which is changing the way both BI practitioners and clients use dashboards. Just like from farm to table, Databricks has some key capabilities to query and build dashboards directly from source to end user, no extra tools required. This AI tool, for example, allows BI practitioners to query data, build visualisations, and understand the data, all through natural language interaction. No coding is required. Genie can even study the data and prompt certain questions or insights you might want to explore. This is truly a game-changer when delivering actionable insights fast.

Databricks AI/BI Dashboards also provide similar features, which allow us to create visualisations through natural language. Once our data is uploaded or directly sourced, we can get automatic suggestions and generations of visualisations that best tell the data story to our end user. But it doesn’t stop there. Once a dashboard is built, through the Genie function and conversational analytics, users can simply ask questions regarding the visualisations and get instant feedback. Compared to traditional dashboards, we now have a dynamic tool that users interact with and derive insights on a whole new level, beyond a standard PPT deck or a meeting with an analyst.

Conclusion

These AI features enable speed and efficiency in the BI workflows, allowing data to be consumed more quickly than ever. Even more so, it frees up capacity for BI practitioners to focus on the more complex tasks, where experience and expertise come into play. Whilst it automates certain tasks like wireframe developments and visual generation, we get a deep dive into building robust data models, form complex logics and calculations, and ensure that the data tells the correct story to the end user. Therefore, it was never AI vs BI or one replacing the other; rather, AI enabling efficient BI workflows as we have never seen before.