Hacker News Reader

Comment by montenegrohugo | Hacker News Reader

I start tons of projects, and it's always a bother naming them. I didn't find existing domain generators at all useful, and since my background is in AI, I made my own.

→ https://www.namy.ai

It currently has a modest but pretty consistent 200-300 users daily, almost all of it direct traffic (my SEO skills are very lacking). I'm assuming people recommend it to their friends, and that's where the traffic is coming from.

It's not yet at $500/mo, but it's getting close. Server costs are significant though, since running an AI model is a bit expensive.

Ideas and feedback are welcome.

View on HN

"Only show available" doesn't seem to work except on the homepage... but I think it's because the homepage is the only place where any domains are actually marked as already being registered. (when most of the suggested domains on search results seem to be registered already, based on a quick sampling.)

On the same line of thought, it would be awesome (but probably difficult/expensive) if you could show the price of each domain directly in the results.

Otherwise, it seems like a neat tool!

You guys are overloading the domain checking API :(

Good problem to have I suppose ^^

That is fantastic, I wish I knew about it earlier. I used another popular name-finding site (can't remember what it was) but it wasn't nearly as intelligent and the results were not that good. It also would be great to check for the availability of the name in Twitter/YouTube/etc.

Looks really cool. The first two results I clicked on where registered a long time ago, though.

https://www.namecheap.com/domains/registration/results/?doma...

Thanks! Yeah, I did NOT expect HN's traffic. It's overloading the domain checking API and making it fail

which is the domain checking API you use? looking for a cheap but reliable domain checking api :)

I think GoDaddy's one is used by a lot of sites?

This is fantastic! If I may be crass, how does this make money? Just through referral links?

Not crass, I like sharing! We're all here to learn from each other.

The monetization model is just referral links to Namecheap, where I get a 10% commission. I want to make that a bit more elegant (especially for people with uBlock Origin, which it doesn't track), and also add a few other referrals (logo makers and maybe hosting).

Couldn't think of other ways to monetize this without making it obnoxious (I hate ads, and making it pay-to-use also seems restrictive to me). If you have any ideas, I'd be open to hear them!

you could try being your own nameserver...through namesilo or something...it's like pretty cheap there already per domain, I think if you used their api as a reseller, and upcharged like 15% on each domain's price you could make more than the namecheap commission...

Thanks! I was impressed to see the revenue figures without any ads on the site (I have similar feelings to you regarding ads and monetisation).

Buy a server with a good graphics card, your server costs will equal out in 3-6 months and then you will actually have revenue.

Cool, I made one called https://mixmatchdomains.com but it never really got much traffic (single digits per day). Maybe the AI thing makes it more marketable, or just being able to type in an open textbox for pure simplicity.

Works surprisingly well.

Thanks! Was a fun project to do. I have a bunch of ideas to make it better, but I decided to let it rest for a bit and focus on other stuff. Might put in a bit more effort if it keeps getting the interest it's getting now!

This is great! I tried "chicken burgers for aliens"* and it came up with some fantastic options.

* not currently a real business plan

Wait until we make first contact (I'm sure this is scheduled later in 2022)

Why are the server cost high ? Can you not run inference on CPU ? Perhaps on an hertzner auctioned machine ?

I’m getting a lot of false positives but this has already generated some great ideas!

I'm sorry! I didn't expect the HN hug of death. About 200 concurrent people right now, so the domain check API is failing :(

Love it. Explainability would be a nice feature. I.e word definitions, origins, etc.

great site, well made. congrats

Have you thought about deploying the ML model on the edge, using something like TensorflowJS or ONNX-runtime?

Haven't done it myself (looking into it right now!), but my impression is that model quantization (and possibly prunning) can give you a palatable model size that doesn't affect performance too much.

Reply on news.ycombinator.com