Pervasive, Data-Driven Decisions: The Promise of Integrating BI and AI
by Saurabh Abhyankar, Executive Vice President & Chief Product Officer, MicroStrategy
It’s long been a C-suite dream to enable data-driven decision-making not just among senior executives, but for all employees. After all, sales reps working with high-value prospects and customer service agents dealing with customer issues all make daily decisions that affect the organization’s health. If these frontline employees had easy access to relevant, reliable real-time information, they could base their decisions not on their gut feelings, but on hard data.
The advantages of making data-driven decisions are clear. For example, McKinsey found that companies using data-driven B2B sales growth engines saw above-market profit growth of 15% to 25%. Imagine expanding those kinds of gains to every department inside the organization. Unfortunately, however, extending access to all employees has proved an almost impossible task.
Also Read: Extreme Networks and Intel Join Forces to Drive AI-Centric Product Innovation
The first hurdle is fragmented data. Critical information is siloed in different applications and locations throughout the organization, making it very difficult to get a holistic view of it all, and CIOs are well aware of the problem. For example, when Foundry surveyed CIOs for its 2024 State of the CIO report, they asked what their priorities were for becoming more data-driven. The No. 3 response was to make their data more available. But data transformation projects that unify data and break down silos can take years and millions of dollars to complete. Those organizations that can figure out how to 1) provide actionable data to frontline workers on demand and 2) empower pervasive data-driven decision-making will gain a significant competitive advantage over their competitors.
Challenges to enabling enterprise-wide data-driven decisions
One might ask why business intelligence platforms don’t fit the bill. After all, these platforms should already be tied into most data sources within an organization, so the lack of a successful data transformation and unification effort shouldn’t be an issue. Unfortunately, access to traditional BI is not simple to scale. The typical approach to enterprise BI is focused on dashboards, and creating dashboards can be a time-intensive task. Designing one for every frontline role would likely result in a long backlog for business analysts, and the dynamic nature of timely, contextual data needed at the frontlines would quickly become a maintenance nightmare.
What’s more, while it’s possible to drill down into the dashboard to get additional information, doing so requires some training. Frontline workers are making lots of rapid decisions daily — if an insurance agent realizes on a call that she needs a specific piece of information about the customer to sell a policy, she doesn’t often have time to browse through multiple dashboards to find it. She needs that data immediately.
Generative AI (GenAI) would appear to offer a solution, since users can interact with it in natural language. Certainly, deploying generative AI is a high priority for enterprises. Just over 60% of companies surveyed this year by Bain & Company said that generative AI is a top three business priority, and nearly 90% said they were already working on deployment, development or a pilot. But GenAI, alone, cannot solve the problem of enabling pervasive data-driven decisions. GenAI is, at its core, a pattern recognition and prediction engine, not a data analytics platform.
So, while it’s very good at parsing natural language and providing answers in ordinary text that sound right, GenAI’s responses may not always be 100% factually accurate and that goes double for numbers and calculations. GenAI is a language model, so it wasn’t designed to perform mathematical analysis. Certainly, AI researchers are improving models to reduce the instances of GenAI “hallucinations,” but as it stands now, GenAI alone is not a reliable source of business information on which to base important decisions.
Better together: BI + AI
There is a way forward that doesn’t depend on an extended data unification project. When GenAI is wrapped around a BI engine that provides trustworthy insights from data, GenAI can do what it excels at — parse human language and communicating in a way that is more appropriate for most users. Organizations can ensure that GenAI doesn’t produce answers with fabricated figures, and users can get the information they need without flipping through multiple dashboards. Instead, users can simply ask for information using ordinary language.
There are many applications for the combination of AI and BI within the enterprise, but one model for rolling out access to insights to all users involves creating an overlay that works with the organization’s standard web browser, which is the interface through which users access most, if not all, of the applications they use. In this way, organizations can provide relevant data to employees within their daily workflow. For instance, one could envision a no-code interface that highlights key words — customer, SKUs, employee names and so on — and produces a card with key data points about that topic when the cursor hovers above them. With GenAI, employees could ask for more detail or for different information entirely, and because it is grounded in BI data, the responses will be trustworthy.
Also Read: More than 500 AI Models Run Optimized on Intel Core Ultra Processors
However the solutions are architected, the combination of BI and AI enable the enterprise to go beyond destination dashboards built by data analysts to empower frontline workers to request the information they need in their own words. We’re on the cusp of realizing the vision for a truly data-driven enterprise. Together, AI and BI will make it a reality.
[To share your insights with us as part of editorial or sponsored content, please write to psen@itechseries.com]
Comments are closed.