Building AI-Tools: Why I Spent 3 Months Cataloging 125+ AI Tools and What I Actually Learned

Honestly, I didn't expect this project to turn into what it became. Three months ago, I just wanted a simple list of AI tools I'd actually tried. I kept forgetting which ones I liked, which ones were worth paying for, and which ones I should never touch again.

So I started writing them down. One became ten became fifty became 125+ AI tools cataloged with detailed reviews, ratings, and actual usage notes. And along the way, I learned some things that absolutely nobody tells you about the AI tool ecosystem.

Let me save you three months of your life.

The Backstory: Why Even Bother?

I've been building AI-focused side projects for about two years. Every week, I'd see another "10 AI Tools You Must Try This Month" article pop up in my feed. I'd check them out, get excited, try one or two, and... nothing. Most of them were either:

• Just wrappers around OpenAI with a fancy UI
• Solving problems I didn't actually have
• Horribly overpriced for what you get
• Already dead by the time I tried them

So here's the thing: I got tired of clicking through Twitter threads and Medium articles just to find out if a tool was actually worth my time. I wanted one place where I could answer the question:

Would I actually recommend this tool to a friend?

That's how AI-Tools started. It's open source, community-driven, and every tool has actually been tested by me (and now other contributors). No affiliate links, no sponsored posts, just honest opinions.

What I Built

The project is pretty straightforward:

• 125+ AI tools categorized by use case (writing, coding, design, productivity, etc.)
• Each tool has: description, pricing, pros, cons, my personal rating, and notes
• Simple JSON format that's easy to parse and reuse
• Automatic validation via GitHub Actions so data stays clean
• Searchable through the repo and can be easily imported into your own projects

Here's what an entry looks like:

{
  "name": "ChatGPT",
  "category": "chat",
  "pricing": "free+$20/month",
  "rating": 4.5,
  "pros": [
    "Reliable",
    "Wide ecosystem",
    "Plugins extend functionality"
  ],
  "cons": [
    "Sometimes slow during peak times",
    "Knowledge cutoff",
    "Can be verbose when you want brevity"
  ],
  "notes": "My daily driver for most tasks. Still the gold standard.",
  "url": "https://chat.openai.com"
}

You can literally drop this into your own project and build whatever you want on top of it. Need an AI tool recommendation engine? Done. Want to build a browser extension that suggests alternatives? Done.

The Real Lessons: What Nobody Tells You

Okay, here's the good stuff. After testing 125+ AI tools, what did I actually learn that you won't see in those "best of" lists?

1. Most AI tools are just hanging fruit

Like, seriously. 80% of AI tools solve the exact same problem that's already been solved by ten other tools. The only difference is the UI and the marketing budget.

I lost count of how many "AI note-taking" apps I tested. Every single one promises to "connect your notes with AI" and "bring your knowledge to life." And do you know what most of them do? They chunk your notes, embed them, store them in a vector database, and let you chat with them. That's it.

Don't get me wrong — that's useful! But it's also the same thing everyone else is building. The space is crowded, and differentiation is hard.

2. You only actually use a handful of tools regularly

Here's a statistic that might surprise you: after cataloging 125+ tools, I probably use less than 10% of them on a monthly basis. The rest I tried once, wrote up my notes, and never went back.

Why? Because most of us don't need that many AI tools. You need:

• A good chat AI (ChatGPT / Claude / Gemini)
• A good coding AI (GitHub Copilot / Cursor)
• A good image generator (MidJourney / DALL-E)
• Maybe one or two specialty tools for your specific use case

That's it. Anything beyond that is just novelty. You don't need it.

3. Price doesn't correlate with quality

This one blew my mind. Some of the best AI tools I found are completely free, and some of the most expensive ones are absolute garbage.

Case in point: I tried a $50/month AI writing assistant that couldn't outperform free ChatGPT. Meanwhile, there are free tools like Ollama that let you run powerful models locally on your machine — completely free, and it's amazing.

Another example: MidJourney is $10/month for basic usage and still produces some of the best images out there. Some of its $50/month competitors can't touch its quality.

The takeaway: Don't pay for something just because it's expensive. Try the free alternatives first. You might be surprised.

4. Maintenance is a real problem

Here's what nobody talks about: AI tools die all the time. I started this project three months ago, and already 7 tools I initially added are either shut down, changed their pricing model dramatically, or just stopped working.

The AI space moves that fast. Startups pivot, funding dries up, founders lose interest. If you're building something that depends on third-party AI tools, you need to be ready for things to break.

That's actually one of the reasons I open-sourced this catalog — the community can help keep it updated as things change. Pull requests are always welcome!

5. The best problems are boring

The most successful AI tools I've seen aren't the ones doing crazy AGI stuff. They're the ones solving specific, boring problems that actual humans have.

Want an example? Notion AI isn't the most advanced AI out there, but it lives where you already take notes, and it solves the exact problem of helping you write better notes. That's it. And it's wildly successful because of it.

Another example: GitHub Copilot doesn't build entire applications for you. It just autocomplete your code. Boring problem, massive value.

Stop trying to build the next AGI. Solve one specific boring problem really well, and you'll have a product people actually pay for.

Pros and Cons of This Approach

Should you do something like this? Let me break it down honestly.

Pros

1. You develop an eye for what's actually good — After testing 125 tools, you can spot a quality product within five minutes. That intuition is worth its weight in gold.

2. It's a great conversation starter — People love talking about AI tools, and this project has opened so many doors for me to connect with other builders.

3. The community actually wants this — I've been surprised by how many people have starred the repo and contributed. Clearly, I wasn't the only one tired of low-quality "best AI tools" lists.

4. It's a good side project that doesn't take over your life — You can incrementally add tools whenever you have time. No big launch pressure, just steady progress.

Cons

1. It takes way longer than you think — Reviewing one tool properly takes 15-30 minutes. Multiply that by 125, and you're looking at 30-40 hours just for testing and writing. That's a lot of spare time.

2. Keeping it maintained is ongoing work — As I mentioned earlier, tools die, change, get acquired. This project isn't "done" — it needs constant updates.

3. It's hard to monetize — I made it open source because I didn't want to do affiliate links or sponsored content. That keeps it honest, but it also means it doesn't make any money. That's fine for a side project, but worth noting.

4. You end up with decision paralysis — Ironically, after cataloging all these tools, I sometimes find it harder to choose which one to use. More options isn't always better.

How You Can Use It

The project is right here on GitHub — feel free to star it, fork it, contribute. Here are some ideas for what you can do with it:

1. Browse it yourself when you're looking for an AI tool for a specific job
2. Contribute your own reviews for tools I haven't added yet
3. Use the JSON data to build your own projects on top of it
4. Add it to your own project as a dependency if you're building an AI tool directory

Every entry is just a JSON object, so it's dead simple to work with. Here's a quick example of how you could use it with Node.js:

const fs = require('fs');

// Read all tools
const tools = JSON.parse(fs.readFileSync('./tools/all-tools.json', 'utf8'));

// Find all coding tools with rating >= 4.0
const topCodingTools = tools
  .filter(t => t.category === 'coding' && t.rating >= 4.0)
  .sort((a, b) => b.rating - a.rating);

console.log(topCodingTools);

That's it. No complicated APIs, no authentication, just pure data.

Closing Thoughts

Three months ago, I started this project because I was frustrated with the state of AI tool discovery. I wanted a single, honest source of truth that wasn't trying to sell me anything.

Along the way, I learned more about the AI ecosystem than I expected. The biggest lesson? Most AI tools aren't as useful as they claim to be, and the best ones are often the simplest.

If you're on the fence about building something like this — just do it. Even if nobody else uses it, the learning experience alone is worth it. I went into this thinking I'd just make a list, and I came out with a much clearer understanding of how the AI tool world actually works.

What Do You Think?

I'm curious — what's the most surprising thing you've learned testing AI tools? Have you found any hidden gems that don't get enough attention? Drop a comment below and let me know!

And if you use the catalog, feel free to star it on GitHub — it helps other people find it. Contributions are always welcome too.

Happy hunting for your next favorite AI tool! 🚀