Skip to content
Articles
IntermezzoAI translation· 30 June 2025

Intermezzo 3 🇩🇪 Trial, Error, Pay, Repeat

by 3 min read

Well, turns out it ain't that simple. That was the gist of the

of “it's promp.td”. Rolling out AI tools in a company without thinking it through can quickly end in a belly flop if you let the hype do the driving and fall into the “Simplicity Trap”.

Now even the McKinsey crowd* have caught on:


"Growth in the generative AI era looks like a classic case of “two steps forward, one step back.” As companies come to grips with the unique complexities of gen AI, initial progress leads to reversals and redos, in some cases threatening to stop development altogether.

There are many sources of frustration and delay, from a lack of sufficient talent to ongoing data quality issues. But our experience working closely with more than 150 companies on their gen AI programs over the course of two years reveals that two hurdles along the building journey almost always surface:

  • Failure to innovate: Process constraints, lack of focus, and cycles of rework that quash innovation. Teams that could be solving valuable problems are stuck re-creating experiments or waiting on compliance teams, who themselves are struggling to keep up with the pace of development. In our experience, roughly 30 to 50 percent of a team’s “innovation” time with gen AI is spent on making the solution compliant or waiting for their organizations’ compliance requirements to solidify and be practical. Teams work on problems that don’t matter, duplicate work, and create one-off solutions that can’t be reused and often fail to unlock real value.
  • Failure to scale: Risk concerns and cost overruns that choke off scale. For the few solutions that show real value potential, enterprises largely fail to cross the chasm from prototype to production. Security and risk concerns (including reputational risk) when scaling gen AI applications are handled individually and become too large and expensive to overcome."

I have no idea how much McKinsey got paid for these insights. I'd happily settle for 1% of it 😀 Fine, trial, error and repeat will probably get you somewhere too. But do you really need such pricey consulting for that? Shouldn't good consulting spare you the learning curve, or at least shrink it to the bare minimum?

Even better are their “proven strategies for overcoming them”:

McKinsey chart "Gen AI reference architecture": a densely packed diagram with dozens of boxes and arrows covering the platform portal, shared data platform, use cases, cloud models, data science, compliance automation, shared services and governance processes.
McKinsey & Company

Seriously? Yet another dreadful consultant-waffle graphic (the text is every bit as muddled as the graphic) that throws everything into one pot and that probably not even its creator understands anymore.


Okay, enough bashing. Integrating this technological innovation really isn't that easy. It's simply different from the IT tools we've known and used for years. You won't get around a certain amount of “trial & error”. But if you go about it the right way, you can climb the learning curve faster and gain a competitive edge. A few recommendations on that can be found in Episode 10.6 - Generate or delegate?


*Here's the paper. Maybe you'll have more fun with it than I did: Overcoming two issues that are sinking gen AI programs


Live long and prosper 😉🖖


mscuibzeug 🤔

▶Listen