With budgets now tightening across corporate America, and the era of easy money a fast-fading memory, the time is nigh for achieving a long-sought goal in the world of business intelligence and analytics: closing the loop.
As far back as 2001, at data warehousing firms like my old haunt of Daman Consulting, we touted the value of “operationalizing” business intelligence. The idea was to leverage BI-derived insights within operational systems dynamically, and thus directly improve performance.
Though embedded analytics have been around for decades, it’s fair to say that most BI solutions in this millennium have focused on the dashboard paradigm: delivering high-level visual insights to executives via data warehousing, to facilitate informed decision-making.
But humans are slow, much slower than an AI algorithm in the cloud. In the time it takes for a seasoned professional to make one decision, AI can ask thousands of questions, get just as many answers, and then winnow them down to an array of targeted, executed optimizations.
That’s the domain of applied intelligence, a closed-loop approach to traditional data analytics. The goal is to fuse several key capabilities – data ingest, management, enrichment, analysis and decisioning – into one marshaling area for designing and deploying algorithms.
There are many benefits to this approach: transparency, efficiency, accountability; and most importantly in today’s market? Agility. During times of great disruption, organizations must have the ability to pivot quickly. And when those decisions are baked in via automation? All the better.
It also helps in the crucial domain of explainability, the capacity to articulate how an artificial intelligence model came to its conclusion. How explainable is a particular decision to grant a mortgage loan? How repeatable? What are the biases inherent in the models, in the data? Is the decision defensible?
On a related topic: The AI Market: An Overview
Take It To the Bank
The rise of fintech startups and neobanks, coupled with rapidly changing interest rates, has put tremendous pressure on traditional financial market leaders to innovate rapidly but safely. Rather than embrace a rear-guard strategy, many firms are looking to AI to regain momentum.
As CPTO for FICO, Bill Waid has overseen a wide range of banking innovations. UBS reduced card fraud by 74%, while Mastercard optimized fraud detection in several key ways, including automated messaging to solve the omni-channel conundrum of communications.
The Mastercard story demonstrates how a large financial institution is now able to dynamically identify, monitor, and manage client interactions across a whole host of channels – and fast enough to prevent money loss. A nice side benefit? Less-annoyed customers.
In a recent radio interview, Waid explained another situation where collaboration improves marketing. “In banking, from a risk perspective, one of the most profitable products is credit card. So if you were to ask somebody from risk: which would you push, it would be the credit card.”
But other departments may disagree. “If you ask the marketing person, they have all the stats and the numbers about the uptake, and they might tell you no, it’s not the credit card, at least not for this group (of customers), because they’re actually looking for a HELOC or an auto loan.”
The point is that you can drive away business by making the wrong suggestion. Without collaborating around common capabilities from a centralized platform, Waid says, that mistake would have likely gone into production, hurting customer loyalty and revenue.
With an applied intelligence platform, he says, key stakeholders from across the business all have their fingers in the pie. This helps ensure continuity and engagement, while also providing a shared baseline for efficacy and accountability.
Think of it as a human operating system for enterprise intelligence, one that’s connected to corporate data, predictive models, and decision workflows, thus achieving cohesion for key operational systems. In the ideal scenario, it’s like a fully functioning cockpit for the enterprise.
This transparency leads to confidence, a cornerstone of quality decision outcomes: “That confidence comes in two dimensions,” he says. “The first is: can you understand what the machine is doing? Do you have confidence that you know why it came to that prediction?
“The second element is that in order for the analytic to be useful, it’s gotta get out of the lab. And many times, I see that the analytic comes after the operationalization of a process, where there is more data, or a flow of data that’s well warranted to an analytic.”
For more information, also see: Best Data Analytics Tools
Bottom Line: The Analytic Becomes an Augmentation
This is where rubber meets road for applied intelligence: the analytic becomes an augmentation. And when the business has that transparency, they get comfortable, and they adopt the insight into their own operational workflow. That’s when the intended value out of the machine is felt.
“Platforms provide unification: bringing process, people, and tech together,” Waid says. And as AI evolves, with Large Language Models and quantum computing closing in, it’s fair to say that the practices of applied intelligence will provide critical stability, along with meaningful insights.
Also see: 100+ Top AI Companies 2023