9/
Code, evals and notes are all here:
https://t.co/Q0xjcES3o6
Live app:
https://t.co/gIhXI9HFrr
Would genuinely love feedback, especially if you can find a claim it gets wrong.
we keep noticing the same thing while building with AI:
it’s incredibly good at sounding right, even when it quietly makes up product details. That feels especially dangerous for launch copy, docs and AI-generated product answers.
I built a tool to test that problem.
🧵
id love to hear anyones suggestions on this ☝️
some other things that broke along the way were background jobs timed out, search struggled with specific questions on long docs, one index setting silently failed when the name got too long
so many annoying edge cases around the AI
@KeugeneLee@squilly Will check it out! id love to get ur feedback on https://t.co/VuP5jIYOWT too - it helps students + newgrads prove what they can actually do through projects,GitHub,startup work & other real evidence. then turns that into a clearer skill profile & matches w relevant opportunities.
@serhiichernenko@workmark We’re still pretty early, but the product is live and we’re actively iterating based on user feedback. Rn we’re focused on improving the profiles, proof-of-work signals and matching between candidates and opportunities. Would love if u checked it out @ https://t.co/VuP5jIZmMr