I agree that it's delusional to replace a company with more than 30000 employees with a bit of LLM and a single motivated dev. Yet, with the rapid progress over the last two years, I do see a realistic risk for standard software in general.

Adobe software can cover thousands of use cases with all their complixities and in all combinations. A single customer usually doesn't need that full set, but only a fraction of these features. So far, it was the rational decision to just buy a capable software off the shelf because development, maintenance, migration etc. for a custom solution was too expensive compared to the license fees. And there I'm quite sure that it will change in the future. Building something tailor-made is getting cheaper and cheaper and might result in more efficient workflows.