A sceptic's guide to tools that solve problems

Let me guess: You're tired of hearing about AI.

Every tool claims it'll change your life. Every newsletter won't shut up about "breakthroughs." Every meeting, someone brings up how AI is going to fix everything. 

And you're just sitting there thinking: Can something just work? Without all the noise?

Yes. Me too. And we’re not wrong to feel this way.

Because, most AI tools don't actually make you faster. There's a study that examined experienced developers' use of AI tools. They took 19% longer to finish their work. Not faster. Slower. But they thought they were faster.

At the same time, people's trust in AI dropped by 18% in 2025, even as more people used it. And when researchers asked 6,000 CEOs about AI's impact, most said it had little impact.

The hype is everywhere. Of that we’re sure.
The results? Well, not so much.

But let’s not be all negative. Because some tools work. You just have to know which ones pass the skeptic test. 

What I actually look for

Before I even consider an AI tool, I ask myself:

Does it fix something that actually annoys me? Not some made-up problem. A real thing that wastes my time every day.
Would I care about this if it wasn't called "AI"? If the answer is no, it's just hype with a label.
Does it make things simpler or more complicated? Some tools need so much babysitting they create more work than they save.
Can I trust what it gives me? If I have to check everything anyway, what's the point?
Can I just use it without thinking about it? The best tools don't feel like "AI tools." They just feel like tools.

If something fails any of these, I don't bother.

The tools I use (and why)

AI for research

I don't trust AI to just tell me things. I need to see where it got the information.

Perplexity shows you the sources for every answer. It's not perfect, it still gets things wrong sometimes, but at least you can check.

NotebookLM is different. You upload your own files, and it only answers from those. No making stuff up. No pulling from random internet corners. Just your documents, made searchable.

Why I like them: I stay in control. The AI isn't the expert, my sources are.

AI for the boring stuff you have to do anyway

This is where AI actually helps. Not by doing your thinking. Just by handling the tedious parts.

Otter.ai records meetings and writes up what was said. You'd take notes anyway. This just does it for you. You still decide what matters.

Granola does something similar but works with different meeting apps. It captures who said what and pulls out the action items. Simple and useful.

Grammarly has been doing this for years, you probably didn't even think of it as "AI." It catches typos, suggests better phrasing, tells you when something sounds off. Does it replace editing? No. Does it catch stuff you miss when you're tired? Yes.

Why I use them: They do one thing well. They don't try to "change how you work." They just remove the annoying parts.

AI tools for making your current tools less annoying

Some tools feel clunky because they added AI later, like an afterthought. The good ones were built with AI from the start.

Cursor (if you write code) rebuilt the whole editor around working with AI. Instead of adding AI features to an existing tool, they designed everything to work together. The research backs this up, tools built this way actually help with big refactoring tasks in ways add-ons can't.

Superhuman sorts your email so you see what actually needs your attention first. It doesn't write emails for you. It just figures out what matters and puts that at the top. Less time searching, more time dealing with things that count.

Why I like them: They cut down on jumping between things. They don't make you adapt to the AI—they shape the AI around how you already work.

The AI tools you don't even notice

Yes. The best AI tools don't feel like AI at all.

Spotify recommendations have been using this stuff for years. You don't think about it. You just get music you like.

Google photos search lets you type "photos of dogs" and it finds them. It just works.

Gmail's smart compose finishes sentences if you let it. Some people hate it. Some people forget it's even there and use it all the time.

Why they're good: You don't have to think about them. They just make what you're already doing a bit easier.

The ones I avoid

Bad AI tools are pretty easy to spot once you know what to look for.

They promise you'll be 10x faster. You won't be. The research is pretty clear on this, most people get about 10% faster, not 10x. If someone's promising you'll become superhuman overnight, they're either lying or they haven't actually used their own product.

They make you rebuild your whole workflow. I've seen this happen so many times. You download a tool, and suddenly you're reorganizing everything to accommodate it. That's backwards. Good tools should fit into what you already do. If you're spending a week restructuring your entire process just to use a new app, that's not efficiency, that's busywork.

They won't tell you how they work. Look, I get it, some AI stuff is complex. But if a company can't explain in simple terms what their tool actually does, that's a red flag. "It uses advanced algorithms" isn't an answer. If it's a black box and you're just supposed to trust it, walk away. You're not asking for the source code. You're asking for basic transparency.

You spend more time crafting the perfect prompt than just doing the task. This drives me crazy. I've tried tools where I spent 20 minutes writing and rewriting prompts to get something that would've taken me 5 minutes to just do myself. That's not a tool. That's a chore with extra steps. If using something requires you to become an expert in talking to it, it's broken.

They talk more about being "AI-powered" than actually solving your problem. I read a piece recently where someone described these tools perfectly, they feel like AI was just stuck on top of something that already existed. Like someone took a normal app, sprinkled some AI on it, and called it revolutionary. Nobody asked for it. Nobody needed it. But now it's there, taking up space in the interface, and you can't turn it off.

What I actually think

AI tools won't replace you. They won't make you 10x faster. They won't change your life.

But they will handle boring stuff if you let them. They will mess up sometimes, and that's okay. The good ones just feel like regular tools.

And here's what the data shows: most companies are still figuring this out. Most AI stuff is still in the testing phase. The hype is way ahead of what's actually happening.

I think what frustrates people most is the gap between the promise and the reality. You hear all this talk about AI transforming everything, and then you try a tool and it... saves you 10 minutes a week. Maybe. That's not nothing, but it's also not the revolution everyone keeps talking about. And that disconnect makes people either over-believe or completely tune out. Neither is helpful.

So if you're skeptical, good. You're paying attention. The trick isn't using every AI tool that launches. It's finding the couple that fix problems you already have.

How I test new AI tools

If you want to try something without getting sucked into the hype:

Always start with the free version. If there isn't one, that's already weird. Why would a tool that's supposedly so useful not let you try it first? Either they don't trust their own product, or they're banking on you forgetting to cancel a subscription. Neither is a good sign.

One at a time. Not five. Not ten. One. You're trying to see if it actually helps, not building a collection. I've watched people download six AI tools in a week and then wonder why they're overwhelmed. Test one thing properly before moving on.

Use it for two weeks. For real work. Not just playing around with it to see what it does. Use it the way you'd use it if you kept it forever. Because the first day, everything feels exciting and new. By day fourteen, you'll know if it's actually useful or just shiny.

If you forget about it, delete it. This is the real test. The good ones become part of your routine so naturally you stop noticing them. The bad ones sit there demanding your attention, sending you notifications, reminding you they exist. If a tool needs to constantly remind you to use it, it's not solving a real problem.

If you miss it when it's gone, keep it. That's honestly the only thing that matters. Not whether it's "AI-powered" or whether everyone else is using it. If you uninstall something and two days later you're annoyed it's not there anymore, reinstall it. That feeling, that little moment of "damn, I wish I still had that thing", that's how you know it was working.

Conclsuion

At Solveo, we test a lot of AI tools. Most of them don't work. Most are just hype.

But a few, a small numbe, really fix problems and saves a lot of time.

Those are the ones I use. Not because they're "powered by AI." But because they work.

And if you hate AI tools? That's completely fine. Just use the ones that don't feel like AI tools.

Because the best tools are the ones you stop noticing.

Keep reading all about AI on our blog! 

‍