Quick build-in-public post, plus a thesis I'd genuinely like people here to poke holes in.
For years I made a living from informational content sites. Over the last couple of years I've watched AI-generated answers quietly take the click. When a search result can summarize your page above the link, a page whose entire value is "knowing the answer" stops earning the visit.
So I've been trying to reason about what actually survives this, and I keep coming back to one distinction:
Answers — the value is in knowing something. Summarizable by definition, which means displaceable. At structural risk.
Transformations — the value is in doing something to input the user brings. Nothing to summarize in advance, because the input doesn't exist until they paste it in.
"How many words should an essay be" is an answer. "Count the words in this draft I just wrote" is a transformation. AI can pre-answer the first. It can't pre-answer the second, because the second needs your data to exist at all.
I'm now testing this with a small project — kitset.app — a shelf of single-purpose, client-side tools (word counter, JSON formatter, case converter, more coming). No accounts, no uploads. Each one is deliberately boring, and each one needs the user's own input to do anything.
The caveat I keep tripping over, in the spirit of not overselling: passing the "can't be summarized" test is not the same as being defensible. A single tool is trivially clonable — by competitors and by the AI itself. Whatever moat exists lives at the level of the whole collection and the trust around it, not any individual tool. I don't have that part solved.
So, two honest questions for this group:
Is the answers-vs-transformations line as durable as I think it is, or am I overfitting to my own situation because I needed a reason to pivot?
For those of you running tools or utilities — are you actually seeing that traffic hold up better than content right now? Or is that wishful thinking on my end?
Would rather have the holes poked now than in six months.