The Illusion of Innovation in the Age of AI
The Illusion of Innovation in the Age of AI
by Pablo Castillo
Everywhere we turn, there are warnings about artificial intelligence. Analysts, consultants, and thought leaders remind us that before we can use AI meaningfully, we must have our data in order. It is a refrain that repeats across articles, panels, and boardrooms.
And yet, the pull of AI is hard to resist. The tools are exciting. They make us feel clever. Hours can disappear experimenting with image generators, rewriting ideas in the cadence of a tabloid headline, or stripping polite but stale greetings from emails so they feel freshly personal. These amusements are real, but they also raise questions: What are we actually making? Are we generating new ideas, or recycling our own through a different lens?
If AI writes code faster, is that innovation or simply acceleration? If it gives us new insights, are they truly new, or are they reflections from the vast library it has already absorbed? The distinction matters, because it shapes how we define progress.
Climbing With the Team
My name is Pablo, and my career has been built inside some of the largest companies in the world. I began by moving early Access databases that supported help desks into Salesforce, back when the AppExchange was just getting started and the industry was shifting toward the cloud. Later, I helped lead massive efforts involving millions of records, setting rules and governance that determined how data would be used.
I have scars from that work. I have seen projects collapse for lack of funding, systems too complex to untangle, and teams pushed past their limits. Success in those environments is not about lone heroics. It is more like climbing Mt. Everest. It is not enough to reach the summit. You must bring your team back down safely. In corporate life, it is the same. Success means more than personal glory. It means ensuring the entire team endures and succeeds together.
Which is why today’s AI race feels familiar. Too many want the shortcut, the plane to the summit. No training, no preparation, no conditioning, just speed. But the cost will come due. In a year or two, we will see the difference between those who planted orchards and tended them carefully and those who cut down the tree for quick fruit.
Where the Work Really Begins
History has a way of repeating itself. Companies chase the newest tool, abandon legacy systems, and leap toward the latest promise without addressing the basics. I see it every time I join a new team.
The first step is always the same: take inventory. Catalog every application in use. What emerges is often chaos, hundreds of tools scattered across the business, each carrying fragments of data. No governance. No clear ownership.
From there comes the task of categorization. Finance, billing, real estate, marketing. Where is the data coming from? Who owns it? How is it being used? Just as important, who will champion the effort? Without leadership’s support, no initiative moves forward. Identifying a subject matter expert is as essential as mapping the systems themselves.
This is the groundwork. Then comes the harder task: imagining how these systems talk to each other. Finance data feeding billing, billing connecting to marketing, all through shared models. If none exist, you build them. You tag data. You break apart dashboards and study the questions they answer. Then you rebuild them, grounded in cleaner, more resilient models.
At the same time, you put structure around the work. Create a charter. Document the vision, the current state, the future state, the risks, and the partners who will help. This charter becomes a compass. When resistance comes, and it always does, you can point back to the shared agreement.
The Real Climb
This is when the real climb begins. You find duplicate pipelines costing millions, teams duplicating effort, and proposals for new systems that serve little purpose other than convenience. Saying no becomes just as important as saying yes.
One of the first exercises is to define the essentials in each area. Finance GL, for example, is anchored by the Chart of Accounts, Journal Entries, Debits and Credits, Trial Balance, Subsidiary Ledgers, Posting, Adjustments and Closing Entries, and the Financial Statements Link. Start there. Have a plan before engaging subject matter experts, because you will encounter systems born from acquisitions, neglected integrations, and data universes that feel foreign even within the same company.
The long-term goal is the creation of gold datasets, trusted sources for finance, billing, real estate, and beyond. These must be carefully mapped, designed to connect with one another, and able to support whatever models the company chooses to build.
It is not glamorous work. It requires persistence, testing, troubleshooting, and often rebuilding. I call myself a system whisperer because success depends less on brute force than on listening carefully to where the system resists and responding with patience.
Asking the Right Questions
This is the conditioning required to climb your company’s Everest. Which is why it is worth pausing when someone presents an AI solution.
Ask: Are we creating something truly new, or is this a problem that robotic process automation could have solved years ago? Is our data mature enough to carry us as a team, or are we leaving people behind? Can this be explained simply, to both the call center agent and the senior engineer? Can it be broken down into terms anyone can grasp? And finally: can you show me the code?
Because in the end, the difference between illusion and innovation is not just speed or novelty. It is the work beneath the surface, the slow and deliberate climb that ensures everyone makes it up and back down the mountain together.
Comments
Post a Comment