GenAI Is Transforming Business Intelligence With Easier Access to Insights

The continuing rise of generative artificial intelligence has the potential to transform the way people work in a myriad of ways, and one of the most intriguing possibilities lies in how it can help to make business intelligence software more accessible, enabling true “self-service” access to data-based insights.

The fusion of generative AI with BI has already taken on a name, “Generative BI,” a concept that promises to transform the way in which BI software analyzes and interprets business data. By circumventing the need for carefully coded queries, which has traditionally been the realm of data science specialists, Generative BI is leading us to a new era in which every business worker can independently leverage their company’s data for strategic decision-making.

In a recent interview with TechBullion, Pyramid Analytics CEO and Co-founder Omri Kohl explained that Generative BI has the power to enhance how data analysis models are created. Following a February update to Pyramid’s platform, users can now use spoken prompts in any language to ask the LLM of their choice questions both complex and oversimplified, initiating “conversations” that yield insights as well as dynamic, interactive data visualizations.

Indeed, the integration of AI gives BI tools the ability to generate more predictive insights, unearth more nuanced data-driven narratives and simulate outcomes. It’s a significant change that will shift the BI industry from passive data analytics to an era of more proactive storytelling.

Delivering on the Promise of Self-service BI

According to Kohl, one of the most important advances of Generative BI is that it will pave the way towards an era of true self-service business intelligence, which is something that has long been promised, but never quite realized.

“Self-service BI never really delivered on its promise,” Kohl said, noting that users continue to grapple with the same data-related problems that made their lives impossible prior to the rise of BI itself. He explained that those issues have actually “metastasized” because the amount of data people interact with through BI has ballooned rapidly.

“What’s more, the idea of self-service created new problems, like multiple versions of the truth, because now everyone has their own copy of the data,” he added.

This explains why the adoption of BI remains low in many businesses. According to research by Gartner, BI software is still only used by around 29% of employees at most organizations, and the percentage of front-line workers actively using such tools has remained more or less flat for the last seven years.

Gartner cites a number of reasons for the low adoption of BI, with one of the main ones being that data preparation remains a complex, tedious and manual process that requires specialist skills. BI tools can’t perform this task themselves, resulting in a major bottleneck.

There’s also a distinct knowledge gap. While BI software can help workers to navigate reports and dashboards without any coding skills, those users still need help preparing and setting up those environments without AI. Beyond the access control and database query skill bottlenecks, proper self-service requires an understanding of concepts such as the underlying key performance indicators and business logic metric definitions, which can be tough for many people to grasp.

Kohl said that the main issue for most business users is that they don’t generally think in terms of pie charts or scatter plots. “They don’t actually know what visualizations or formats to apply to their data, but self-service BI only works if you have some level of expertise in data manipulation and exploration,” he said. “People really just want a simple system that delivers answers to their questions.”

Generative BI allows users to ask questions of their data in their natural language, and generate a dashboard or a pre-sliced chart that provides them with the answers they’re looking for in a way they can easily understand. “With Generative BI, BI is now truly self-service because anyone can ask a question in natural language, without any expertise in data exploration, and receive the insights they need in under a minute,” Kohl asserted.

Shifting from Descriptive to Prescriptive Insights

People in virtually any business environment can appreciate the power of data to support better decision-making that’s based on solid facts and evidence. But the BI industry has struggled to provide people with a way to access those all-important insights that are locked within their data in a timely way.

“Until now, business leaders didn’t always have the expertise to analyze data and produce meaningful insights, and if they had the tools and the capabilities, they couldn’t get to those insights in time to make tactical decisions at the speed of business,” he said.

Generative BI is making it possible for line-of-business workers without any specialist skills to come in and ask a question in their everyday language. The system can decode the intent behind that question, identify the most suitable data sources required to answer it, and work out the most effective data query that needs to be performed on those datasets. It can also leverage the correct statistical analysis framework for the task at had. All of this is done under the hood, Kohl explained, so the user only receives an answer to their question that’s in an easily digestible format.

The reliance on predefined dashboards and reports will diminish, resulting in a shift of data analytical power from BI analysts to line-of-business workers, freeing up both to spend more time on higher value work.

Business users will be better able to self-serve themselves, asking questions and getting the insights they need without assistance. In turn, that will give BI analysts more time to focus on higher-value tasks, such as ensuring data quality, managing database life cycles, and creating a semantic layer that documents the entire catalog of knowledge trapped within the company’s data.

With this capability, Kohl argues that Pyramid Analytics is paving the way for a more nuanced shift that will see data being used in less of a descriptive manner, and more a prescriptive way, empowering users to make more strategic and tactical decisions.

“We’ve already come a long way towards bridging the gap to prescriptive data management by simply delivering actionable answers to business questions, in under a minute,” Kohl said. “We’re bringing the same kind of automated magic to corporate data that Google Maps brought to journey-planning. In the same way that you enter your destination into your navigational app and follow the instructions, we’re hoping that Generative BI will soon be able to provide strategic advice based on data trends and projections.”

Smarter Applications

By embedding Generative BI into their builds, individual applications will also be able to leverage Generative BI, Kohl said, resulting in a new generation of more intelligent apps that can provide smart recommendations, with or without being prompted.

For instance, if a user asks a banking app how they can save more money, it can provide them with various tips based on an analysis of their spending habits and investments.

But those intelligent apps won’t necessarily require a prompt to be helpful. “A financial advisor might deliver automated smart suggestions about how clients can optimize their portfolios,” Kohl explained. “Or manufacturing factory management apps might provide generative, interactive supply chain efficiency insights.”

Pyramid Analytics has also set its sights on evolving Generative BI to overcome challenges around “scalable personalization,” which isn’t very practical at present due to privacy concerns. Kohl said people want to be able to ask very specific and pointed questions about personalized entities.

While this is not possible currently because the underlying large language models require detailed data to respond accurately to such queries, Kohl is confident that Pyramid will have a viable solution in place soon enough. “We’re exploring ways to get over that hump of situation-specific data investigation, and we’re very close to success,” he said.