Frameworks / Different Verticals, Same Wall
Different Verticals, Same Wall
Manufacturing, automotive, healthcare, retail and banking describe their AI challenges as though they were industry-specific. They are not. Each arrives at the same orchestration wall from a different direction.
Sit in enough of these conversations across enough industries, as I have across Asia Pacific, and a pattern becomes hard to unsee. Every sector believes its problem is unique to it. The manufacturer talks about shop-floor data and machine reliability. The bank talks about risk models and regulatory exposure. The hospital talks about clinical validation and patient safety. The language could not be more different. The wall behind the language is the same.
The wall is orchestration, not the domain
The domain-specific part of any AI problem is real, but it is rarely where things break. What breaks is orchestration: getting fragmented data to move, getting models to run reliably against real systems, getting the organisation to actually change how it works, and sustaining all of that past the initial enthusiasm. Those challenges wear industry costume, but underneath they are structurally identical.
This matters because the vertical framing does active harm. It convinces each industry that it must solve the problem from scratch, that lessons from other sectors do not apply, and that the specialists who understand its domain are the ones to lead — when the binding constraint is orchestration expertise that transfers cleanly across all of them. The framing hides exactly the help that would work.
How to see it in the field
The tell is a leadership team convinced that no one outside their industry could possibly understand their problem. When “our sector is different” becomes the reason not to look at how other sectors solved the same orchestration challenge, the vertical framing has become a wall of its own — one that keeps the useful lessons out.
A second tell: the programme is staffed almost entirely by domain experts and thin on people who have moved AI into production anywhere, in any industry. Deep domain knowledge is necessary but not where the failure lives. If the team has no one who has hit the orchestration wall before, it will hit it as though for the first time.
Getting out of it
Separate the two problems explicitly. Name the part of your challenge that is genuinely domain-specific — the clinical validation, the regulatory constraint, the physical process — and treat it as exactly that: specialist work. Then name the orchestration part, and go looking, deliberately, for how organisations in completely different industries solved it. The cross-industry pattern is where the leverage is.
The most useful benchmark for a bank stuck at the wall is rarely another bank. It is a manufacturer or a retailer that hit the identical wall two years earlier and got past it. Same wall, earlier. That is a teacher the vertical framing would never have let you see.
Part of the Frameworks series. Related: the Pilot Trap, the Technology-First Trap and the Solo Trap.