Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
alexy201
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
4 ms
·
1.
▲
Show HN: Public and Free SWE-bench-lite evaluations
(swebenchevals.com)
3 points
by
alexy201
2y ago
|
1 comments
2.
▲
Show HN: Accountability as an AI
(github.com)
1 points
by
alexy201
2y ago
|
1 comments
3.
▲
by
alexy201
2y ago
I’ve been finding it hard to push myself to complete my New Year’s resolution of going to the gym and getting jacked. So, I built an AI agent that yells at me and keeps me accountable. Introducing Goggins AI. Inspired by @anuatluru’s #Work
4.
▲
by
alexy201
2y ago
Absolutely agree, although from a business perspective it is a good strategic move, it just kills innovation—especially for users that DO want to try out cool features or new products. Like, I can't even give MY OWN data to a service t
5.
▲
by
alexy201
2y ago
Would love to hear your guy's thoughts on this. And again, excuse my language in the document.
6.
▲
A 19-year-old's commentary on LinkedIn services
(docs.google.com)
2 points
by
alexy201
2y ago
|
3 comments
7.
▲
by
alexy201
2y ago
isn't this the same as GPT 4o?
8.
▲
by
alexy201
2y ago
I like the minimalist design of this tool, looks like it can be really useful!
9.
▲
by
alexy201
2y ago
Really cool stuff, this looks like it can be useful.
10.
▲
by
alexy201
3y ago
Sorry about that should be fixed! Try on Desktop, it might work better
11.
▲
by
alexy201
3y ago
Yeah a little bit?
12.
▲
by
alexy201
3y ago
Also here's a fun little story I cooked up with AI: https://nsfwgod.com/history?historyID=b1089628-f2d5-4298-bbe...
13.
▲
by
alexy201
3y ago
Made by a 19 year old—give me your worst.
14.
▲
Roast My Winter Project
(nsfwgod.com)
2 points
by
alexy201
3y ago
|
6 comments
15.
▲
by
alexy201
3y ago
The person that designed these icons should be paid less. Maybe Bezos did it himself
16.
▲
by
alexy201
3y ago
Hey everyone, I am the creator of AWS Docs GPT and its been extremely useful to garner all of your feedback for the site—thank you guys so much! We are constantly improving and updating the GPT, including less hallucinations, more accurate