A point from René Sellmann made me think a bit differently about software moats.
I used to put quite a lot of weight on being the “system of record”. I still think it matters, but maybe not as much as I thought.
AI doesn’t actually need to replace Adobe, Intuit, Autodesk, etc. for the economics to get worse.
The software can remain sticky because the data still lives there, while an AI layer starts doing more of the actual workflow. Customers might still renew, but you could still get fewer seats, less direct usage or weaker pricing power.
So I’m trying to separate a few things more clearly:
* where the data lives
* who owns the workflow
* how hard the underlying data is to recreate
* who still gets paid if the interface changes
For me the interesting part is that retention could still look quite good while the moat is already getting weaker economically.
This has changed how I think about a few of my holdings. I’ll probably go through them individually later.
I made a simple diagram of what I mean here:
[https://substack.com/@michaelhillaert1/note/c-351471578](https://substack.com/@michaelhillaert1/note/c-351471578)
