Kit Cox is Founder and CTO of Enate, a process orchestration and AI solution built for B2B services.
It’s been nearly three years since ChatGPT burst onto the scene. AI has been heralded as a productivity savior in boardrooms, become the keynote speech at conferences and even dominates the conversation at social gatherings.
But while hefty investments have been made and adoption has been widespread, the returns are lackluster at best. A recent MIT study found that 95% of businesses are getting zero ROI from AI. Zip. Nada. Tumbleweed. AI fatigue has well and truly set in.
It’s not that the technology isn’t powerful; it is. It’s just that AI in isolation won’t fix broken operations. Without a holistic plan for improving your business, AI becomes about as useful as a widget in the corner of your browser.
Things need to change, or we’re looking at another RPA nightmare in the making.
AI Dreams Vs. Reality
Walk into any business meeting today, and someone will inevitably start waxing lyrical about the need for AI adoption. They’ll tell you that AI can draft proposals, write code or solve communication problems in minutes. On a personal level, these tools are easy to plug into daily life and extract value from. They’re intuitive, responsive and useful for everything from researching holidays to conjuring recipe ideas with leftovers from the fridge.
But when applied in the business sense, things get complicated. The MIT study revealed a difficult truth. Despite $30 billion to $40 billion in enterprise investment into generative AI, just 5% of organizations are seeing profit and loss benefits. Researchers have dubbed this the “GenAI Divide”—a Grand Canyon separating companies extracting millions in value from the vast majority stuck with no quantifiable impact.
Over 300 public AI initiatives were analyzed. For quick personal assistant-style tasks like writing emails or briefings, AI was preferred. However, add any layer of complexity, and 90% prefer to work with a human.
A lawyer in the report admitted: “It repeats the same mistakes and requires extensive context input for each session. For high-stakes work, I need a system that accumulates knowledge and improves over time.”
Most Business Operations Are Fundamentally Broken
While AI is far from perfect, there’s an elephant in the room that has nothing to do with technology. In most conversations I have with business leaders, it becomes apparent that their operations are fundamentally broken.
Most businesses can’t answer the most basic questions: How much work is being done in your operation? Who’s doing it? What are your bottlenecks? Where do you need to add or remove resources? Add to that the fact that most businesses rely on emails and spreadsheets to manage complex work, and you have the problem. When you try to bring AI into this environment, it’s no wonder it fails.
MIT researchers solidified this point: Most AI implementations fail “due to brittle workflows, lack of contextual learning and misalignment with day-to-day operations.”
This mirrors the hype cycle we saw with RPA a decade ago. Grand promises, isolated use cases and no way to scale because the underlying processes weren’t ready. Many companies spent millions on RPA implementations that delivered minimal value because they automated broken operations rather than fixing them first.
Orchestration Is The Bedrock Of AI
Before you can successfully deploy AI at scale, you need to create the visibility and structure that AI requires to be effective. If your business involves delivering services at scale, you probably need process orchestration.
What is orchestration? It’s one interface to manage all work in your operation across people, systems and processes. Think of it as a musical orchestra where everyone plays their part at the right moment to create the right outcome. One single layer where you can see all work in real time, identify where processes need improvement and manage resources effectively.
Orchestration is the bedrock of any automation strategy. With the visibility and data it provides, you can track performance accurately and identify specific tasks where AI will have the biggest impact. More importantly, you create a structured environment where AI can actually learn from outcomes and get better over time.
Here are a couple of tips for getting started with orchestration:
Implement orchestration gradually.
Start orchestrating with one well-defined problem. Pick a workflow, pilot orchestrating it, get it right, then roll it out. Organizations that try to overhaul everything at once run into problems.
The biggest piece of advice I can give for a successful orchestration deployment is to bring your team along from day one. Open communication is vital, and citizen development—where people closest to the work help shape how it’s orchestrated—is one of the most effective ways to manage change. When employees feel part of the design process, adoption sticks.
Another stumbling block is vendor choice. I’ve seen implementations drag on for two years because businesses bought heavyweight systems that needed endless custom development. Look for a solution that fits around your existing tech stack, goes live in months rather than years and flexes as your needs change.
Get your house in order, then adopt AI.
Once you’ve got the visibility and data that orchestration brings, you can start seriously thinking about automation.
The move to AI should be deliberate and incremental. As with orchestration, start small, be clear on what you want to achieve and build from there. Put governance and guardrails in place to protect quality and compliance, then introduce AI into targeted processes where it can deliver quick wins. An onboarding process, for example, is full of repetitive, high-volume work.
Two examples I’ve seen deliver real value are: sentiment analysis, which is a way of monitoring customer comms in real time and flagging customer frustration before it becomes a churn issue; and email classification, which can route thousands of messages instantly and give hours back to service teams.
Conclusion
AI fatigue is real, but it’s not inevitable. The 5% of organizations seeing returns are the ones that take the bigger picture seriously. They build solid foundations, get their house in order and then apply AI strategically. Everyone else will keep wondering why their investments don’t pay off.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?