Generative AI in analytics: Enhancing the user experience without overreach
As scepticism creeps more and more into the generative AI discussion, I’ve been trying to gain a clearer view of where real value lies, particularly for tools used directly by end users. I was therefore in the right frame of mind during a recent briefing with Karel Callens, CEO of Luzmo, who offered some interesting and useful insights into AI’s role in analytics. The discussion also prompted some broader thoughts on how generative AI is being integrated into the analytics platforms in general.
Enhancing Analytics: A Growing Trend
Before diving into the specifics of our conversation, it’s worth noting that the integration of AI to enhance analytics user experiences is a growing trend across the industry. Major solutions like Tableau, Microsoft Power BI, and IBM Cognos (plus many others) are all incorporating AI features such as natural language querying, automated insights, and intelligent recommendations. This broader context underscores that while Luzmo’s focus on embedded analytics for SaaS providers gives the company a unique angle, the general direction of using AI to improve user interaction with data is widely recognised as valuable.
On Luzmo itself, its specialism is enabling SaaS vendors to build analytics features into their solutions, a space that’s evolving rapidly as user expectations rise and technology advances. “Our primary focus is on analytics for client-facing applications,” Callens explained. “AI enhances our core offering, rather than replacing it.”
This made a lot of sense to me and other colleagues on the call. Instead of positioning AI as replacing traditional front ends, Callens clearly described it as a tool to augment existing point-and-click interaction models that still work well for many users in many scenarios. This approach alleviates concerns relating to reliability, repeatability, explainability, and trust that some might have about total reliance on generative AI for analytics across the board.
Practical Applications and Future Potential
Coming back to Luzmo, one typical use case Callens described was an AI-powered chart generator. “We’ve made this available as an additional feature, and about 20% of users are actively engaging with it at the moment,” he shared. This approach suggests a thoughtful implementation, allowing users to adopt AI-enhanced features at their own pace.
An idea I hadn’t really considered before is AI’s potential ability to enhance analytics personalization. Callens explained, “AI opens up possibilities for hyper-personalised insights. We can now consider not just the user’s context, but also factors like the time of day and current data trends to deliver more relevant analytics.”
This level of contextual awareness could be really useful for non-analyst users providing more relevant, timely insights to inform decisions without having to define environment and role based parameters explicitly. Importantly, Callens was careful not to overpromise. “While we’re making significant progress, we’re not quite there yet in some areas”.
But on the use of natural language, Callens confirmed what we’re hearing more broadly, that “The conversational aspect of AI is becoming more important,” adding “It allows for continuous refinement of insights through user interaction, which in turn helps improve the AI’s performance over time.” He did reiterate, however, that natural language can coexist with traditional interfaces if you get the UI design right, offering users choice in how they interact with their data.
A Balanced Approach to AI in Analytics
The lingering thought I was left with following the discussion with Callens was that when it comes to generative AI in analytics, context and balance are crucial. In this domain, AI has the potential to significantly enhance user experiences and deliver more personalised insights. However, it’s about augmenting existing tools and human expertise, not replacing them entirely.
The approach described by Callens, and reflected in the broader industry trends, suggests that you don’t have to go all-in on AI to reap benefits. Using generative AI as a front-end to enhance the user experience can deliver substantial value, and this doesn’t mean taking away familiar interaction mechanisms that many users will find more accessible or, indeed, more efficient.
From a personal perspective, I love natural language interfaces, but not for everything in all circumstances. Plus I have also come across quite a few users who really struggle with them, especially knowing where to start when presented with a blank box. That’s why I took to Callens’ idea of using natural language to refine something you are already looking at – “Show me that by region” or “Change it from units to percentages”, etc.
To be honest, with all this talk of AI changing everything, I still think one of the interesting parts of the discussion is around user interaction, so I will continue exploring this area until things shake out a little bit.