Agents act without you clicking. That changes what software is for. Features won’t save products that can now be skipped entirely.
The story
Patterns that make this sound AI-written:
– Repetitive structure: short punchy lines stacked in the same rhythm, over and over
– Generic contrasts: “this vs that” framing without lived examples
– Abstract language: “systems,” “workflows,” “category-level change” with no concrete scene
– Predictable transitions: “This isn’t a hypothetical,” “The uncomfortable part is”
– Overconfident claims with no texture: reads like a keynote, not a person thinking out loud
– No personality leakage: no self-doubt, no small human moments, no quirks
– Clean, polished phrasing: no friction, no rough edges, no “how did I learn this?”
Rewritten version:
What happens when software stops waiting for you to show up?
Most software runs on a quiet deal. You open it, click around, it helps. You’re driving, it’s the car.
Agents mess with that deal.
This isn’t smarter autocomplete. Picture something checking your stuff, making a call, and taking action while you’re still deciding if you want coffee or water. You didn’t press a button. It already moved.
Now I start wondering something a bit uncomfortable. If an agent can *do* the task, what job is left for the app that used to help you do it?
I used to think better tools would keep winning. Build faster, add features, done. Then I watched people spin up an agent to track a project instead of opening a project tool at all. Same outcome, fewer steps. Huh.
The old safety blanket for a lot of mid-level software was “this is hard to build.” That blanket has holes now.
So what sticks around? Things with real pull – network effects, data no one else has, rules that make switching painful. The rest feels… replaceable. Not dead. But sweating.
And “more features” does not fix this. That’s like adding cup holders while the car learns to drive itself.
If you’re thinking about your roadmap and feeling a bit of that stomach drop, yeah, same. I’ve b
Recent Comments