Quick Takes: AI Tools I Actually Use Daily

I don’t use most AI tools. They solve problems I don’t have. But a few have actually embedded themselves into my workflow, not because they’re flashy, but because they save me real time on work I actually need to do.

Claude is the obvious one. I use it for thinking through half-formed ideas, catching logical gaps, and handling the tedious parts of writing: outlines, rewrites, finding the right word when I’m stuck. It’s not writing for me; it’s a conversation partner who doesn’t get tired. I’ve also started using it for code review and debugging, which is honestly where it shines most. It catches the stupid mistakes I’d otherwise waste an hour hunting for.

Copilot in VS Code is the thing I didn’t think I needed until I used it. It’s not going to architect your system for you, but it’s relentlessly good at completing obvious code patterns, which means fewer keystrokes and fewer moments of “what’s the syntax for this again?” I turn it off for the tricky parts and keep it on for the repetitive bits. That’s the move.

For research, I keep coming back to Perplexity. It’s a search engine that reasons through results instead of just listing ten links. I ask it something half-formed, and it actually synthesizes an answer. Works especially well for “explain this trend” or “what’s changed in this field since last year?” It’s not perfect, but it’s a time saver when you’re trying to get context fast.

I also use Figma’s AI features more than I expected. I’m not a designer, but when I need to spin up a quick mockup or layout, the auto-generation features get me 70% of the way there. Beats starting from scratch and then tweaking. Same energy as Copilot. It’s doing the grunt work, leaving the judgment calls to me.

The common thread: all of these work because I’m using them to augment work I already know how to do, not replace it. I know how to write. I know how to code. I know how to design a basic interface. These tools just remove the friction on the parts that are mechanical. That’s the filter. If a tool makes you feel like it’s doing something you can’t evaluate, that’s usually a sign you shouldn’t trust it yet. This is why I wrote AI won’t replace you, but someone using AI might. The risk isn’t the tool itself, it’s falling asleep while others get faster.

I wrote about this before in how I actually use AI in my daily workflow, and it’s worth revisiting: the tools that stick are the ones that respect your expertise instead of trying to replace it. You might also want to check out what I stopped delegating to AI. Knowing what these tools shouldn’t do is just as important as knowing what they should.