For about a year, I used ChatGPT for everything. Writing, research, brainstorming, summarizing documents, drafting emails, trying to understand my health insurance explanation of benefits. Everything. It was just easier to have one tab open and one mental model of what I was working with.
I think a lot of people do this. You find a tool that works, you get comfortable with it, and the idea of learning another one feels like more trouble than it's worth. ChatGPT was good enough. Why fix what isn't broken?
Then a few things happened that made me start testing alternatives — and I realized I'd been leaving a lot on the table.
I had to write a difficult email. Someone I'd worked with for years had been consistently dropping the ball on a shared project, and I needed to address it directly without blowing up a relationship I wanted to keep. I asked ChatGPT to help. The output was fine. Professional, inoffensive, covered the main points.
A colleague mentioned she'd been using Claude for writing. I figured I'd try it with the same prompt, mostly out of curiosity. The difference was noticeable enough that I actually stopped and read it twice. The Claude version felt like something a thoughtful person would actually write. It understood the emotional nuance of the situation — the thing I was trying to do was be honest without being cruel, and it got that.
I sent the Claude version.
A few weeks later, I needed to dig into some information about a topic where the details mattered and being wrong would have been embarrassing. I asked ChatGPT, got a confident-sounding answer, and used it. I later found out one of the key claims was just... off. Not wildly wrong, but wrong enough that I had to go back and correct it.
I started using Perplexity for anything factual. Every answer comes with citations you can actually click. If something is wrong or outdated, you can see where it came from and judge for yourself. The experience of using it for research is so different from ChatGPT that it's almost a different category of tool.
I've settled into a workflow that uses four tools. This sounds like more overhead than it is — each tool has a clear job, and I rarely have to think about which one to reach for.
The honest version of why most people stick with one AI tool isn't that it's the best one for everything — it's that switching has a cost. You have to learn the new interface, rebuild your mental model of what it's good at, and accept that some prompts that work in ChatGPT need tweaking in Claude.
That cost is real. But in my experience it's about a week of adjustment before the new tool stops feeling foreign. And after that week, the upgrade in output quality for the right tasks is worth it.
Don't start with the tool everyone uses. Start with the task you most need help with, and find the tool that's actually best for it.
If you write a lot — proposals, emails, anything client-facing — try Claude. If you do research and accuracy matters — try Perplexity. If you need to brainstorm or generate a lot of variations fast — ChatGPT is still excellent at that. If you just want something free and surprisingly capable for everyday questions — DeepSeek.
You don't have to use four tools. Even switching one thing — using Perplexity instead of ChatGPT for research, or Claude instead of ChatGPT for important writing — is a meaningful upgrade.
One last thing: the reason I built this site is that I kept having these "wait, this tool is so much better for this" moments and wanted somewhere to track the comparisons properly. If you're figuring out which AI actually fits your workflow, the quiz below is a faster version of what took me about six months to work out on my own.
Answer 5 questions and get a personalized recommendation based on how you actually work.
🎯 Take the Quiz →