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

Let me start with a confession: I have a problem.

Every week, I see 10 new AI tools tweeted, posted, and ProductHunt'ed. "This one will change your life!" "10x your productivity!" "The only AI tool you'll ever need!" And every week, I click through, sign up for the free tier, play with it for 20 minutes, and... forget about it.

Sound familiar?

Honestly, I got sick of it. So three months ago, I decided to do something crazy: I'd catalog every AI tool I could find, test it, categorize it, and finally answer the question that's been bugging me: do we really need 1000+ AI tools, or are we all just falling for the same hype over and over again?

Today, I've got 125+ tools cataloged in AI-Tools — and I learned some things that surprised even me.

The Project: What I Actually Built

So here's the thing — I didn't just want a list. Anyone can make a list. I wanted something usable. Something I could actually use when I'm looking for a tool to solve a specific problem.

Here's what I built:

• A Node.js based catalog with structured JSON data for every tool
• Automatic categorization by use case (productivity, development, design, writing, etc.)
• GitHub Actions that keeps everything updated and generates a static page
• Each entry includes pricing, pros, cons, and my personal rating

The code is dead simple — you can literally fork it and make your own list in 10 minutes. Here's the core structure:

// tools.json structure example
{
  "tools": [
    {
      "name": "ChatGPT",
      "url": "https://chat.openai.com",
      "category": ["general", "writing", "coding"],
      "pricing": {
        "free": true,
        "paid": 20,
        "notes": "Plus gives better model"
      },
      "pros": [
        "Mature, reliable",
        "Great at most tasks",
        "Large ecosystem"
      ],
      "cons": [
        "Can be slow at peak times",
        "Privacy concerns with data",
        "Knowledge cutoff"
      ],
      "my_rating": 8
    }
  ]
}

Then there's a simple script that generates a clean HTML page from the JSON:

const fs = require('fs');
const tools = require('./tools.json');

// Group by category
const byCategory = tools.reduce((acc, tool) => {
  tool.category.forEach(cat => {
    if (!acc[cat]) acc[cat] = [];
    acc[cat].push(tool);
  });
  return acc;
}, {});

// Generate HTML from the grouped data...
fs.writeFileSync('./index.html', generateHTML(byCategory));

I know, it's not rocket science. But that's the point. Sometimes the simplest solutions are the best. You don't need a fancy React app with 500 dependencies to curate a list. A static HTML page generated from JSON works perfectly, and it's free to host on GitHub Pages.

What I Learned: The 6 Big Surprises

Okay, let's get to the good stuff. After testing 125+ AI tools, what actually stuck with me?

1. 90% of AI tools are just "hanging fruit" for the same problem

This was the biggest eye-opener. Most new AI tools aren't solving new problems. They're just solving the same 10% over and over again.

How many "AI writing assistants" do we need? How many "AI image generators"? How many "AI code assistants"?

Don't get me wrong — competition is good. But after a while, the marginal improvements get smaller and smaller. The difference between the top 5 AI writing assistants is negligible for 90% of users.

What does that mean for you? Stop chasing every new tool. Master the 3-5 good ones you already have. You're not missing much.

2. You only actually use 5-10 tools regularly

I know this sounds obvious, but let me put numbers to it. After three months of testing, I still use the same 7 tools every single day. That's it. The other 118+? I might check them once a month when I need something specific, but that's it.

My daily 7:

1. Claude - For long documents and complex reasoning
2. ChatGPT - For quick answers and coding help
3. GitHub Copilot - Integrated into my editor, can't live without it
4. MidJourney - When I need images for blog posts
5. Grammarly - Basic proofreading, saves me from stupid mistakes
6. Notion AI - I live in Notion, so it's convenient
7. Perplexity - For research and up-to-date information

That's the whole list. Everything else is niche. And that's okay. You don't need to use everything.

3. Price isn't correlated with quality

I tested everything from free tools to $200/month enterprise tools. And you know what? Some of the best tools I found are completely free. Some of the most expensive ones are... just okay.

For example:

• Free tools that punch way above their weight: Llama 2 running locally, Stable Diffusion, Obsidian with plugins
• Expensive tools that didn't impress me: Several $50+/month "productivity suites" that basically just wrap OpenAI's API and add a fancy UI

Don't get me wrong — I happily pay for good tools. But the "you get what you pay for" rule doesn't hold up as well in the AI space. There's amazing free stuff out there, and there's expensive hype that doesn't deliver.

4. Most "AI-powered" features are just gravy

Here's what I mean: The best AI tools aren't only AI. They're good tools with AI added.

Take Obsidian — it's a great note-taking app first. The AI plugins are nice-to-have, but the core product works perfectly well without AI. Same with VS Code + Copilot — VS Code was great before Copilot, Copilot just makes it better.

The tools that fail are the ones that are only AI. "AI-powered note-taking" that doesn't even have good basic search. "AI-powered task manager" that can't even organize your tasks properly.

AI is a multiplier, not a foundation. Build the foundation first.

5. The ecosystem is still chaotic, but that's okay

Three years into the generative AI boom, it's still the wild west. APIs change, models get updated, companies pivot, free tiers get shut down.

I can't tell you how many tools I added to my list that changed their pricing model or just went away while I was working on this project.

But honestly? That's normal for any new space. The companies that survive will be the ones that focus on solving real problems, not just riding the hype wave.

What does that mean for you as a user? Don't lock yourself into a single ecosystem. Keep your data portable. Don't pay for a year-long subscription to some new AI tool you haven't used for six months.

6. The real value isn't in the tools — it's in the curation

After building this, I get why people keep making these lists. The value isn't in having a huge number of tools — it's in filtering out the noise.

When you're looking for an AI tool for a specific job, do you really want to scroll through 100 options? Or do you want 3-4 solid recommendations from someone who's actually tested them?

That's the secret. Curatorship is more valuable than completeness. I'd rather have a short list of good tools than a long list of every tool.

Pros and Cons of This Whole Experiment

Honestly, was this worth three months of my life? Let's be real about it.

What Went Well ✅

1. I actually know what's out there now — No more FOMO when someone mentions a new AI tool. I've either tested it, or I can check it against my categories.

2. It's actually useful — I use my own list all the time when I need a specific type of tool. Instead of Googling and getting 100 listicles written by people who haven't actually used anything, I just check my own notes.

3. The code is reusable — Anyone can fork this repo and make their own curated list. It's just JSON and a simple generator. No complicated setup.

4. I got better at saying "no" — After testing so many tools, I'm much better at ignoring the hype. If a new tool doesn't solve a problem I actually have, I don't even bother clicking through anymore.

What Didn't Work So Well ❌

1. It takes more maintenance than I thought — Tools change, links break, companies shut down. Keeping everything updated is an ongoing project. I'm still figuring out the best way to handle this.

2. I got distracted by the project itself — Some weeks, I spent more time adding tools to the list than actually using tools to get work done. Meta problem, I know.

3. Rating is hard — Giving a tool a 7 vs an 8 is subjective. What's great for me might be terrible for you. I've learned to focus more on pros and cons than numeric ratings.

4. Discovery is still hard — Even with categorization, finding the exact tool you need isn't always instant. I'm still thinking about better ways to organize everything.

Who Should This Project Interest?

Let me make this simple:

You should fork AI-Tools and make your own list if:

• You're tired of forgetting which tools you liked and which you hated
• You want to curate your own personal collection instead of relying on 10 different listicles
• You like keeping things simple and static (no databases, no hosting bills)
• You want to share your curated list with others

This probably isn't for you if:

• You want a complete directory of every AI tool ever made (there are other sites for that)
• You need a fancy web app with user accounts and search and all that
• You don't care about curation and just want everything in one place

My Advice for Anyone Wanting to Do Something Similar

If you're thinking about doing your own curated AI tools list, here's my advice after three months:

1. Start small — Don't try to add 100 tools in the first week. Add 5-10 a week, actually test them, write real notes. Quality over quantity.

2. Keep it simple — You don't need a fancy tech stack. JSON + a simple generator works perfectly. Don't spend more time building the project than curating.

3. Be honest — If a tool is hyped but you don't like it, say so. Your curation is only valuable if you're honest about the bad stuff along with the good.

4. Update regularly — Set a reminder to go through your list once a month and check that links still work, pricing hasn't changed, and your opinions still hold.

5. Share it — Even if only a few people find it useful, that's still better than keeping it to yourself. I've already gotten great suggestions from people who found my list, and that's made it better.

Wrapping Up

So after three months and 125+ tools, what's the bottom line?

We don't need another huge list of every AI tool ever made. What we need is curated lists from people who've actually used the tools and are willing to tell you what they really think.

Most AI tools aren't changing your life. They're just incrementally improving your workflow. And that's okay. You don't need every improvement. You just need the improvements that work for you.

I learned that I was spending too much time looking for the next great tool and not enough time actually using the great tools I already had. That alone made this whole project worth it for me.

The project is open source on GitHub: https://github.com/kevinten10/AI-Tools — feel free to fork it, star it, use it for your own curation.

Now I'm curious — what's the one AI tool you use every day that I might not have heard of? Drop it in the comments — I'm always adding to my list, and I love finding hidden gems.

Have you ever curated your own AI tools collection? What did you learn from the process?