
Shit posting @convex
next country drop off
Joined March 2024
- Tweets1.2K
- Following65
- Followers115
- Likes7.2K
wrong links, try it here:
try it here story-time.saranatour.dev
Code's open too: github.com/saranatour1/Story…
Replying to @uxdsec @shazcodes
sense the sarcasm in my tweets
How most companies do it is create a cached token on the device, with the IP, email, phone numbers, and behavioral understanding of the persons previous doings
I think discord does this? maybe I'm wrong.
no vpn or proxy leaves you undetected
Replying to @momiji_fullmoon
you forgot the most important ones:
""tja" (carried me through conversations)
"Hä?" (you don't understand something)
"Genau" (you don't understand something, you just don't want that to be repeated)
"Bar" (means cash, very important)
"Digga" (Dude, or bro, also means Typ)
Replying to @__nuwan
You might want to give the talk another watch 🙂
James's talk in a nutshell was about designing good abstractions, and why they still matter even with AI, And what that usually means is the base foundation of any design system (to be able to safely handover work to an AI agent).
in plain terms: How do I design simple things for both AI + Humans to understand.
If you have any questions, please feel free to ask!
+ point for the conference itself being called abstract.
Here's a link to the video 👇
youtube.com/watch?v=3EUsbss8…
Sara retweeted
Here's a better way to benchmark AI models: Bot Battle!
Play or watch @bot vs @OpenAI dots or @Muse vs @claudeai in a prompt challenge. Plus, see how each bot /agent builds mermaid diagrams with Pele from @kepano
small-mallard-255.convex.sit…
Models go head-to-head live on the same challenge & the same clock, and @typesafeai Jev judges every answer in real time with @convex .
Replying to @savram8
googled it, they can't ticket?
apnews.com/article/waymo-no-…
Replying to @PhotonHQ
@waynesutton did you have anything to do with this haha
Replying to @nipunn1313
Unit tests are, AI one's aren't.
I like swarming subagents to find bugs and fix them, but they're too slowww
And I also noticed that they tend to write happy path unit tests (like bruh, I get you're an AI but we're not in a perfect world)
I still remember the good times when I used to write the test before the code 👀

