There’s no shortage of companies talking about how ready they are for AI. Budgets have been approved, platforms have been selected, and employees are being encouraged to find ways to use it. In some organizations, AI has already become part of the daily routine. But I think we’re getting ahead of ourselves when we equate using AI with being ready for it.
I’ve spent much of my career working with businesses on the technology and infrastructure they depend on to operate. One thing I’ve learned is that there’s a big difference between getting something to work and getting it to work reliably across an entire organization. AI is exposing that difference in ways many companies haven’t fully anticipated.
A successful pilot can create a lot of excitement. Maybe a customer service team is responding to inquiries faster, or a finance department has found a way to automate reporting that previously took hours. Those are legitimate improvements, and businesses should be looking for those opportunities. The trouble starts when leadership assumes that because something worked well in a controlled environment, it will work just as well everywhere else. It rarely happens that easily.
Take a company with hundreds of locations, multiple technology providers, and years of accumulated systems and applications. Getting an AI tool to work for one department is relatively straightforward. Making that same tool work consistently across the business is another matter entirely.
Now you’re dealing with different networks, different security requirements, information stored in multiple places, and systems that may not communicate particularly well with one another. Add AI to that environment, and you’re not necessarily fixing those problems. You may be making them more visible.
And when the application becomes something employees or customers depend on, those weaknesses matter a great deal more. I also think we’re overlooking how fragmented responsibility for AI has become.
In many businesses, individual departments are moving forward with their own initiatives. Marketing has its tools. Finance has its applications. Customer service is experimenting with automation, and IT is trying to make sure everything is secure and supported.
Everyone is making progress in their own area, but who is looking at what happens when all of it comes together? Who makes sure the information feeding these systems is accurate? Who determines whether the existing infrastructure can support the additional demand? Who is responsible when an automated process produces the wrong result or a critical integration stops working?
These aren’t hypothetical concerns. They’re the kinds of operational questions businesses have always had to answer when introducing new technology. AI simply raises the stakes because of how quickly it can process information and how much responsibility organizations are beginning to give it. There’s also an assumption that AI will somehow compensate for weaknesses in the systems around it. I don’t see that happening.
An AI application can be remarkably sophisticated and still be limited by an unreliable connection, outdated information, or an integration that doesn’t perform as expected. It can analyze data in seconds, but that doesn’t help much if the data is incomplete or inconsistent.
For companies operating across multiple locations, these challenges can become especially complicated. A business may have invested heavily in cloud applications and digital transformation, yet still be relying on a patchwork of connectivity services, legacy platforms, and processes that were never designed to work together.
Those issues existed long before AI entered the conversation. What’s changing is the level of dependence businesses are placing on technology to make decisions and keep operations moving.
If a reporting application goes down, employees may have to wait for information. If an AI-driven system is responsible for responding to customers, routing requests, or initiating operational changes, the consequences of an interruption can be much more immediate. That should be part of the readiness conversation from the beginning, not something discovered after deployment.
Another area where I think companies need to be more realistic is the financial side of AI. There’s tremendous pressure to show that an investment is producing results. Boards want to see progress. Executives want to demonstrate that their organizations aren’t falling behind. And vendors are understandably eager to show how much time and money their technology can save.
But the cost of an AI initiative doesn’t end with a software subscription. There are integrations to build, systems to maintain, security requirements to address, and employees who need training. Someone has to monitor performance, investigate errors, and make sure the technology continues delivering the results that justified the investment in the first place.
I’ve seen enough technology implementations to know that the initial deployment is often the easiest part. The real work comes afterward, when the system has to function day after day alongside everything else the business is running. That’s why I’m skeptical of measuring AI success primarily by how many employees are using it or how many processes have been automated.
Those numbers may tell you something about adoption, but they don’t necessarily tell you whether the business is operating any better.
Are customers getting better service? Are employees spending less time correcting mistakes? Are processes becoming more reliable? Are operating costs actually improving once the full expense of supporting the technology is considered?
Those are the questions I’d want answered. None of this is an argument for slowing AI innovation. I believe businesses should be experimenting with the technology and looking seriously at where it can create value. The opportunities are significant, and ignoring them would be a mistake.
But there’s a difference between moving quickly and moving without a clear understanding of what you’re building on. Before expanding an AI initiative, leadership should have a realistic picture of the environment it will operate in. That means understanding where information resides, how systems connect, what happens when something fails, and who is accountable for the outcome.
It also means being willing to acknowledge that some parts of the business may not be ready yet. That’s not an admission of failure. It’s a practical assessment that can prevent a much more expensive problem later. I suspect we’ll see a growing divide between companies that have successfully experimented with AI and those that have figured out how to make it a dependable part of their operations.
The difference won’t necessarily come down to who bought the most advanced platform or who moved first. It will have a lot to do with how well organizations understand their existing technology environments, how clearly they’ve assigned responsibility, and whether they’ve invested in the less exciting work required to make everything function together.
AI is moving fast, and businesses have good reasons to be excited about what it can do. But if I were sitting in a boardroom discussing our next major AI investment, I’d want to spend less time hearing about what the technology is capable of and more time understanding what it will take to support it. Because getting AI to work is one thing. Being able to count on it when your business depends on it is something else entirely.
