Is AI actually the hard part of enterprise AI?
retained text
I've spent a lot of time around healthcare and enterprise data, and I'm starting to think the answer is NO The model is often the easiest part of the conversation now. The harder questions are usually much less exciting: Where did this data come from? Can we trust it? Who is allowed to use it? Is the context still there? What happens when the data changes? And can we actually put the AI into a real workflow without creating another mess for the team? I've seen plenty of conversations where everyone wants to talk about the model, but the real work starts when you get underneath the data. That's one reason I've become so interested in the infrastructure around AI, especially in regulated industries like healthcare. What do you think? For people actually deploying AI inside enterprises, what's taking more time right now: the AI itself, or getting the data and systems ready for it?
document fetched 2026-09-24 00:20 · canonical https://reddit.com/r/AI_Agents/comments/1wnvsbk/is_ai_actually_the_hard_part_of_enterprise_ai