Getting ROI from Enterprise AI

Real ROI Starts with Customized Enterprise AI Tools

If you’re reading this, you probably already know what off-the-shelf AI tools can and can’t do…and that they aren’t going to get you the ROI you need. They’re generic solutions trained on generic data, and even when they work well, they’re quickly embraced by competitors.

Real ROI, though, starts with customized gen AI tools. Every organization is a finely tuned system, refined over years to construct and optimize the workflows that generate profit and sustainable value. Gen AI tools can accelerate this, but they must be designed or modified to “understand” those workflows.

That’s why it’s critical to start with desired ROI and work back from there. For a pharma company, ROI comes from the launch of new, more effective drugs. For a retailer, ROI comes from relevant new product offerings and effective marketing campaigns. Starting from this point ensures that a new tool delivers what the organization needs…not just what the technology is good at.

Working back from this starting point, we identify the critical decisions that impact ROI—something we call Human-Centered Workflow Analysis—and develop customized AI tools that support those specific decisions.

Successful tools depend on five basic elements:

Leading Adopters are Using AI to Transform End-to-End Processes

The biggest gen AI success stories we’re seeing right now are where entire processes are rewritten from beginning to end.

Think of filing a claim with your insurance company, or calling your bank about an error: If the experience could be made dramatically faster and more responsive, where everyone you interact with is hyper-informed and supremely capable, that would transform your relationship with the brand behind it. No single tool can do this, but a rethought and restructured process—supported by a customized AI workbench—can.

The e-commerce example is useful here: If you forgot everything you knew (in, say, 1998) about how people bought books, then redesigned the process in the presence of web technology, what would it look like? What tools, structures, and roles would your company need to support this experience?
Transforming a process can be thought of as answering a series of questions:

Taking this approach, a wide range of processes can be made faster, more accurate, more personalized, and more effective. Besides the banking and insurance examples above, many other processes are ripe for transformation. Marketing strategy, for instance, can be reimagined from beginning (user research and actionable insights) to end (marketing and product decisions). A new kind of enterprise innovation could start from AI-enhanced pattern spotting, AI-facilitated scenario exploration, and the testing and evaluation of AI-enabled business prototypes.

Even more than individual customized AI tools, a re-imagined process often requires significant organizational support. This starts with administrative buy-in, and often encompasses re-training, data governance, resource management, and organizational restructuring. The potential for ROI, though, is vast.

AI Agents Will Reshape Industry

The next step beyond AI workbenches is agent-based AI, in which autonomous agents are empowered to seek out information and make decisions on their own, interacting with humans and each other for maximum impact.

Agentic AI is already here and generating value. IA Collaborative is helping clients create and localize agents as well as experimenting with our own.  Each agent challenges how we think about the interface itself. Generative interfaces must now be:

The Forefront is Already Here

Agents are the cutting edge right now, and the organizations implementing them today will be among the first leaders of a new guard. But it won’t be that way for long.

The pace of gen AI adoption is already the fastest ever seen for a new general-purpose technology. Companies who’d never heard of it three years ago are now at the forefront, and new breakthroughs in speed, capability, model size, and affordability happen daily. This is enabling extraordinary new human experiences and business opportunities, which are likely to expand dramatically over the next few years.

For most organizations, the technology isn’t going to be the limiting factor—existing frontier foundation models already have more capacity than most industries can implement. The bottlenecks are in the organizations themselves: the willingness to embrace change, to think about familiar processes in new ways…or to simply know where to start. There’s a re-envisioning that must happen before a company can become truly “AI native”.

So what’s the road to ROI look like then? Because gen AI gains value through customization, every company’s optimal path will be different. What they all have in common, though, is an optimal time to start: now.