Can't disagree more! There's bottomless demand for more software. Here's a few examples I encountered just in the last few weeks that wouldn't be feasible before LLMs:

- More localism. Are you afraid of being cut off from tech by some future US government? Now it's feasible for your local culture to grow its own office suite, operating systems, Active Directory competitor etc. A less interdependent world with more competition does have its advantages.

- The building management company for my apartment sucks. Basic problems go unfixed because they appear to suffer extreme labour shortages and serious problems with flaky labour e.g. employees that just randomly go AWOL in the middle of conversations without bothering to tell anyone. A lot of the work of these employees is actually just coordinating and paying contractors in response to problem reports, something that can now be automated by AI ... but they haven't done it yet.

- I just finished assembling some flatpack furniture. Every time I do this it reminds me why IKEA dominates the market. Other furniture companies give the strong impression they don't usability test their instruction leaflets. This should and could be massively better: AR assistance during the build would be great, AI stress-testing instructions to verify they make sense would be great, AI checking every packet has the right number of components in it would be great. And there are lots of furniture companies out there. They don't all need to use a single SaaS to do this.

+ in general robots will require tons of software/models to make them do tasks usefully, especially as they lack training data.

That's just a few examples of places software could have made my life easier in just the last few weeks.