AI At The Speed Of Committee
Talking to someone working in a big company: "Everyone is really keen to get working with AI, as soon as we've completed our vendor assessment process."
The problem is that the landscape moves faster than the speed of these assessments, especially when the goal is to find, by committee, an option for every department in the company. That will just leave everyone disappointed. Typical enterprise procurement processes weren't meant for such fast-moving technology.
What to do, then? A few points.
There is no use in trying to determine "the best" model. That changes on an almost weekly basis anyway. Here's where you should satisfice, not optimize.
There's also no use in finding "the one" for everyone. Sure, it simplifies procurement and IT, but IT's role is to be an enabler. "We picked a crappy solution that nobody is happy with, but at least it streamlines the support work in our IT department." Doesn't sound like a good strategy to me
What matters most is getting on the bus. You can always tweak and improve once you're getting your hands dirty. No ivory tower vendor assessment committee beats hard-won experience on the ground.
If that all sounds too reckless, identify pilot areas with limited blast radius, by all means, just don't shrink away from it just because you haven't formulated the perfect plan yet.
