5 min

Generative business intelligence, or GenBI, is one of the big new excitements in the world of business optimization. It utilizes generative AI (GenAI) to take BI to the next level, inviting business analysts, decision-makers, and strategists to ask natural-language questions about data and receive meaningful visualizations and easy to understand answers.

In many ways, GenBI delivers on the unfulfilled promise of self-service BI (SSBI) tools. A decade ago, SSBI was trumpeted as a revolution for business strategizing. But most people don’t think about their business in terms of graph design, there was still too much friction involved with the initial data discovery processes that needed to take place before they could generate reports.

As a result, many line-of-business users still ask data scientists to code their queries, which makes it too hard to unlock the true value of their data. With the newer GenBI approach, users don’t need advanced data querying skills to obtain relevant, useful insights from their datasets. They can simply ask questions in their usual wording. Not surprisingly, many business users are embracing GenBI tools such as Tableau, Outerbase, Microsoft Power BI, Pyramid Analytics and Sisense.

One survey reports that over one-third of respondents expect AI integrations to improve their use of BI over the next 12 to 18 months.

But GenBI is far from done and dusted. Some of the leading product rollouts are only available in beta mode, have limited capabilities, or are still under development. There are still more developments and skills in the pipeline, at varying stages of readiness and experimentation. Read on to delve into the exciting ongoing possibilities of GenBI.

AI turning data into stories

Storytelling is the best way to deliver insights and explanations in any context, but turning data into comprehensible, cohesive narratives has always been a challenge.

Automated, AI-powered BI tools have begun to overcome that hurdle, adding summaries in simple language to visualizations and reports, thereby providing context and making it faster and simpler for users to grasp and apply insights.

“Narrative BI represents a pivotal advancement within Generative BI, leveraging advanced AI to transform raw data into compelling narratives,” says Golam Mustafa, Head of IT Applications at Emami Agrotech Limited. “These narratives not only illuminate complex patterns but also drive actionable outcomes across various domains, from marketing analytics to sales forecasting and risk management.”  

AI enabling custom formulae

As business data users become more comfortable manipulating data and using different visualizations and report formats, they want the freedom to choose which formulae and methods are used. This means building LLMs that are versatile and “creative” enough to function as a build-a-model salad bar.

LLMs, for all their capabilities in the field of language, still aren’t very efficient at proactively manipulating numbers. It’s also tough to build a data processing model that can understand user requests on a deep level, divine what the user really wants to calculate, draw upon past interactions, and produce a syntax that is ultimately correct.

AI delivering truly scalable analysis

One of the biggest challenges with corporate data analysis is the sheer scale of the datasets involved. Business users want insights that cross massive and highly complex datasets which are constantly updated, and data volumes are only going to keep growing.

“How do you scale without taking your gigantic Snowflake data lake and making it work instantly, without having to build it again?,” asks Avi Peretz, CTO and co-founder of Pyramid Analytics. “How do you fine tune your LLM every two seconds, when the data is hydrated? That’s the last frontier to cross.”

Innovative GenBI companies are working on using AI to make this scalability a reality. The sticking point is that it’s both unsafe to share your data with LLMs, which have many well-known security vulnerabilities, and impractical. These datasets refresh on a near-constant basis, so it would take far too long to upload all the relevant data to your LLM, even if it could be secured. However, solutions like Peretz’s Pyramid Analytics, which is structured to serve as an effective interface between a company’s databases and an integrated LLM, are on the road to resolving the challenge.

AI unlocking real-time data management

Business data is hardly static, even if analysis methods often are. The challenge is to produce analytics insights that accurately reflect the current reality, not that of five weeks, five days, or even five minutes ago.

Many aspects of business operations could be enhanced and improved with finely-tuned responses based on real time trends.

In many ways, this is a subset of the previous issue. Keeping your LLMs updated with real time or even near-real time data is still out of reach, but some of the best minds are working on it as you read this, so the solution could arrive soon.

AI integrating BI into business operations

As GenBI solutions evolve and mature, we’ll see them absorbed into the broader business ecosystem. Embedding BI analytics capabilities into existing apps will allow insights to surface within the flow of work.

“By seamlessly integrating into existing and future workflows, these systems will enable teams of employees to effortlessly carry out their daily operations and make data-driven decisions in real-time,” predicts Zuzanna Pajorska, marketing manager at Stratoflow.

At the same time, GenBI will become integrated into enterprise systems, in the way that Microsoft Fabric is beginning to implement. Pajorska adds that “The next generation of BI systems will cater to a more extensive user base and establish deeper connections with expansive enterprise systems.”

AI supporting highly specific answers

The new wave of GenBI tools use AI to answer natural-language queries in a highly effective way, but currently, those queries can only be about fixed points like locations and dates. “People are dying to ask very specific questions about specific entities in their database, and get an intelligent answer back. This is the next eureka point,” observes Peretz from Pyramid Analytics. “That doesn’t work well right now, because the LLMs don’t have enough detail about the specifics in your database.”

To take their utility to new heights, the models will need a better “understanding” of what a company’s data is all about – which is, of course, a challenge among humans as well. This level of machine understanding will allow people to ask pointed questions about their businesses, even when they know less about what data is available than the AI does, and still receive an insightful answer.

While some GenBI solutions are already somewhat skilled at making these types of logical inferences, newer iterations will need to solve this challenge in a more satisfying manner.

GenBI has more aces up its sleeve

There are still many hurdles lying in the path to data analysis nirvana, but there are good reasons to expect GenBI to be able to overcome them. Scalability, real time insights, meaningful storytelling, embedded analytics, and more are on their way, thanks to the promising convergence of GenAI and BI.

Asim Rahal is an independent consultant who plans and executes IT security strategies and compliance practices across environments. An incurable evangelist of cloud security, data protection, and cyber risk awareness, Asim has been published in Dark Reading, TechTarget, and InfoQ.