Every week I see the same scene: a company gets excited about AI, builds a nice-looking pilot, shows it on the big screen, and six months later nobody uses it. MIT measured this in 2025: around 95% of AI pilots don't move the company's results at all.
It's not a lack of technology. It's a lack of process. I come from production engineering and Lean Six Sigma. The first thing I learned wasn't how to do something, it was how to ask: what's the number that needs to change? Where's the bottleneck?
Most people do the opposite. They buy the tool, then go looking for the problem. That's when AI turns into an expensive decoration. I work in the right order: define the metric, fix the process, and only then put AI to work on the heavy lifting.
When you flip the order, the number shows up. At Marksell, sales went from 27 to more than 500 leads a month. On the growth teams I've led, we hit 103% of target with an average ROI of 11.6x. None of that started with "which model do we use."
And when a company gets it wrong, it usually aims at the wrong target. MIT itself found that more than half of AI budgets go to sales and marketing, while the biggest returns were sitting in the back office, automating process. IBM backs this up: the gain almost never comes from the model, it comes from embedding AI into a workflow that already exists. The money goes to what shines, not to what pays.
I don't write code as an identity. I direct the AI, lead the team, and answer for the result. That's why what I build ends up in the 5% that survive contact with reality.
