I’ve been noticing something lately… a lot of products are rushing to add “AI features,” but very few of them actually feel useful in real workflows.

Not the flashy demos. I mean the stuff that genuinely saves time or improves decisions.

For example:

• Auto-generated summaries that people still don’t trust.

• AI copilots that interrupt more than they help.

• “Smart” recommendations that feel generic.

It made me question whether we’re building AI for users… or just shipping it to stay relevant.

So I spent some time breaking down what actually makes AI features work in real products vs what just looks good on paper. Wrote my thoughts here:
medium.com/@varsha17ojha/ai-features-useful-real-…

Curious to hear from others here:

• What’s one AI feature you actually rely on today?

• And what’s one that feels completely useless despite the hype?

Would be interesting to see where people draw the line between “cool” and “actually valuable.”