I tried more AI tools this year than I can remember. Most didn't stick.

I’ve been trying a lot of AI products lately—writing assistants, research tools, meeting tools, automation platforms, and AI agents.

The interesting part is that most of them were impressive during the first 10 minutes.

Then I stopped using them.

The pattern I noticed was pretty simple: if a tool requires me to constantly change how I work, remember another workflow, or check whether the AI actually did what I asked, the initial excitement disappears quickly.

The tools that stick are usually the opposite. They fit into something I already do and remove a small but annoying step without creating another task to manage.

That made me start thinking about what actually makes an AI product useful beyond the demo.

For me, I now look at a few things:

• Does it solve a problem I actually have?

• Does it save me time after the learning curve?

• Can I understand or control what the AI is doing?

• Does it fit into my existing workflow?

• Would I still use it if the “AI” label wasn't there?

I’m curious what others have experienced.