bstrama
bstrama on Hacker News
Posted by
bstrama on Sunday October 18, 2026 @05:11PM
from the show-hn dept.
Foundation labs use bundled subscriptions to pool inference costs across dozens of tools. Startup founders don't have such advantage.
As the result, we see a lot of product behind 200/mo paywall. Conversion rates are low, free tiers cost builders a lot and Anthropic kills competition at scale.
Please, share your thoughts on that
Could you share GitHub? Also image background remover doesn't seem to work on my PC (macOS 26.5.1, chrome 149.0.7827.103)
Posted by
bstrama on Tuesday June 16, 2026 @10:45AM
from the show-hn dept.
https://wattfare.com/
Hey HN!I recently built halupedia - a funny site that hit 300k+ users in few weeks and #1 on hacker news. It was amazing project, which costed me $300+.
Since halupedia uses a lot of AI I paid a lot for the bill, fortunately I'm fine thanks to donations and resigning from many planned features.It inspired me to built wattfare - just like BYOK it let's user bring their own AI budget to the...
Agreed. If you have some time, you can find our GitHub on halupedia.com. All PRs welcomed! ;)
Nice! I'm thinking on how can I make it more interactive.
Those author styles could work well, will check that out.
Btw how would you imagine such dispute? What do you think could be the trigger for the article to be regenerated?
Could be interesting direction to discover. The only problem with such implementation is it could take some work to make it cheap and actually well working. And I'm just thinking about the near future of this project.
I really like it, but without organic traffic, at the position we're right now, the moment HN stops showing us at the top, we will loose all the visitor.
And it's not like I'm trying to do a startup out of it. I just very enjoy making something people love! It's first time in my life and it's amazing.
If you have any interesting thought, please leave them here - I'll definitely read it, or visit our discord [link on halupedia ;) ].
there's now a search bar btw
It is instructed to reference A LOT of articles. It just hallucinates all the url. If the url points to already existing article - it's just a coincidence
Here's our source code: https://github.com/BaderBC/halupedia
Used to be a problem - now consistent for new articles ;)
If you have idea how to improve it, I'm all ears ;)
Interesting idea with flagging.
We are considering 2 options:
1. You can generate aricle only if it was previously referenced in previous one
2. Flagging mechanism, now that you brought it up.
Let me know what you think!
Actually interesting response. You can also check out github.
I were just drunk and idea seemed funny. That's the idea behind haha.
But either way can't wait to see google ai overview cite us.
Update: Implemented it. All new articles work that way
Best part - I didn't implement such logic. It just for some reason works that way.
Also now that I think, we store articles in decwntralized cloudflare KV store and access from serverless workers running also on their servers.
That could be the thing behind it being so quick.
Cloudflare workers have 1ms cold start.
Just added comment section :)
UPDATE: Just now, comment section added.
Have a nice time arguing!
Exactly, but I consider adding fake search that could find you ANY article, including not existent ones
Yep, just a react. Also we use gemini 2.5 flash lite, so it's fast, cheap and dumb.
Great suggestion! Will immediately look into that!
Yeah, thought about that, maybe will implement it. Will keep in mind!
For now SSR to feed LLMs' the priority
Tried once, but was useless. Very funny that it had so many text, while Elon is apparently "huge" fan of short and precise communication...
It generates articles only once. So once it's generated, it never perish.
Logic looks like:
If article exist -> show it
If not -> generate and save