Hey IH ๐
Quick intro: I spent most of my career as a Product Analyst with legal background, where I lived for years building things alongside engineers without writing a line of code myself. About three and a half years ago I caught the bug and I started learning to ship things with low-code, then with a lot of AI help, and somewhere along the way I turned into what people now call a "vibe coder." I am just someone with product intuition, a legal brain, and enough patience to argue with an LLM until something works.
This post is about the side project that came out of my own frustration.
Every time I needed a new AI tool โ for writing, for editing video, for cleaning data, for automating a tiny piece of my workflow โ I'd do the same thing:
Open one of those giant AI directories.
Stare at 8,000 tools sorted by "popularity."
Click through 15 of them.
Get distracted by ads.
Eventually pick the one with the prettiest landing page.
Regret it a week later.
I'm a product person. I know what a bad funnel looks like. And honestly? Most AI directories are catalogs pretending to be recommenders. They show you everything and help you with nothing. The cognitive load is on you.
I kept thinking: I don't want a list of 8,000 tools. I want the right one for me, today, for what I'm trying to do.
Pickurai is an AI tool recommender. Not a directory.
You answer 6 short questions โ what you want to do, your budget, your skill level, your preferred interface, a couple of context details โ and in under a minute you get a personalized shortlist of 3โ5 tools that actually fit your case, with reasons for each pick.
No signup. No ads on results. Free.
This was the part I cared about most, because I didn't want it to be "just another GPT wrapper that vomits names":
Hand-curated catalog of 397+ AI tools, across 10 categories. New tools added every week. Pricing, features and integrations kept up to date manually.
A scoring methodology I built myself: every tool in the catalog is scored across several dimensions (price-to-value, ease of use, depth, learning curve, language support, etc.). So matching isn't a vibe โ it's math against your answers.
An Anthropic API assist kicks in when the scoring alone isn't confident enough to rank the top picks. Claude helps disambiguate the edge cases so the final shortlist is actually defensible.
The combo of curated data + deterministic scoring + LLM-as-tiebreaker is what makes the recommendations feel like advice instead of SEO.
I genuinely believe AI is the most powerful social elevator we have right now.
The gap between someone who knows how to use the right AI tool and someone who doesn't is becoming the new literacy gap. And the most painful part is that the people who would benefit most โ freelancers on a tight budget, students, people switching careers, small business owners โ are exactly the ones who can't afford to lose hours testing tools or paying for the wrong subscription.
Pickurai matches recommendations to your budget. If you said "free only," you only get free tools. If you said "20โฌ/month max," you don't see a $200/month enterprise plan. That part isn't a feature โ that's the whole point.
Catalog: 397+ tools, growing weekly.
Stack: vibe-coded with a lot of help from Claude, of course.
Business model: free for users, no ads on results. (I'll figure out monetization later โ I'd rather get the usefulness right first.)
Metrics I care about: completion rate of the 6 questions, click-through to the recommended tool, and "did the user come back."
Roast my onboarding. Try the 6 questions and tell me where you bounced or hesitated.
Tell me the AI tool you wish existed in the catalog โ I'll add it this week.
If you've monetized a free recommender before, I'd love to hear how you thought about the transition without making the product worse.
A lawyer-turned-product-person-turned-vibe-coder built this in his nights. If it saves you one wrong subscription or save some money, it already paid off.
๐ pickurai.ai
Thanks for reading โ happy to answer anything in the comments.
The onboarding framing is sharp โ 6 questions beats scrolling 8,000 tools every time. The completion rate metric you're watching is exactly the right one.
On the monetization question you raised: the move that usually works without breaking the free experience is gating one layer deeper โ not the recommendation, but the comparison. Show the top 3 free, let them unlock the full scored breakdown or side-by-side feature comparison for a one-time fee. The value is already in the scoring methodology you built. That's the part worth paying for.
What's your completion rate sitting at right now?
Thanks for the comment, it is very sharp.
I dont want to charge the customer directly nowadays. When the project get traction I hope we can obtain money from ads in the blog and affiliate from the AIs. The bad point is that just 2% of the tools that we suggest have affiliate program and none of the top performance have it too.
At the moment I just want the tool to be visible.
And related to the completion of the site, I would say it is complete at 80% at the moment.