I was reading a report from Contrary Research this morning titled The Case for Specialized AI.
The report argues that the future isn't about one giant, generic model that does everything. It's about specialisation—using specific, smaller models to handle specific tasks with high precision.
The Gap Between Hype and Reality
What really stood out to me was the data on why businesses are struggling to see returns on their AI spend.
"Across the board, there is a general gap between the hype and realised productivity... 45% of enterprise executives reported returns below expectations."
Why the disappointment? The report notes that generic tools are "easy to demo but hard to build into workflows."
Amazon employees have described outputs from internal generic AI tools as "slop," and studies show that unmanaged AI contributions often increase software bugs rather than fixing them.
The "Upskilling" Trap
This reveals the fundamental flaw in the current strategy.
Right now, many organisations are trying to fix this "slop" problem by training everyone to be an "expert." They are throwing licenses at their accountants and sales teams and hoping they turn into "Prompt Engineers."
But if the tools are "hard to build into workflows," why are we putting that burden on general staff?
Driving the Car vs. Building It
To fix this, we need to understand the difference between driving the car and building it.
1. The Team (The Drivers)
General training is essential, but its goal should be confidence, not construction. Your team needs to know the road rules. They need to feel safe behind the wheel and understand what the "car" is capable of. This prevents fear and sparks curiosity about where they can go next.
2. The Specialist (The Builders)
However, you wouldn't ask your delivery driver to build a new transmission in the parking lot. That leads to breakdowns. Actual implementation—connecting AI to your data, ensuring security, and preventing "slop"—is a job for a specialist.
The "Plain Logic" View
The data shows that specialised agents outperform generalists. The same is true for your operations.
For the Team: Focus on confidence and safety.
For the Build: Bring in a specialist.
Your team shouldn't have to wrestle with buggy code to do their jobs. They should have reliable, specialised vehicles built for them—so they can get back to driving the business forward.