Demo agents look magical, production agents look brittle, where does it usually break for you?

Almost every enterprise AI project we've worked on has the same emotional arc. Demo wows the room. Production reveals everything the demo hid.

Where it usually breaks, in order of frequency I've seen:

- Real customer data is messier than the demo data

- Edge cases the demo never hit start showing up daily

- Compliance review adds 3 to 6 weeks of unplanned work

- Whoever built the demo doesn't have time to operationalise it

What's been the demo-to-production gap that hit you hardest?