The Growing Fatigue Around "AI-First" Products

There’s a growing backlash happening across software right now: people are getting genuinely tired of “AI-first” everything. What was initially novel now often feels disingenuous, homogenous, and stripped of human touch.

First Rule of AI Engineering: Figure out where NOT to use AI. It should be your last resort, not a default feature.

Large language models are fundamentally probabilistic next-token predictors.
Forcing a chat interface onto everyday workflows that demand deterministic, repeatable logic isn’t innovation — it’s lazy UX that adds unnecessary friction.

When a core product relies on an unpredictable guess, and the “guardrail” is just another LLM babysitting the first, users can tell.

You aren’t solving a user problem; you’re engineering user frustration.

The smartest product strategy right now isn’t figuring out where to shove AI next, it’s having the discipline to know where to keep it out.