sdk.vercel.ai

Currently I only use Claude for PDF to cards as it is very convenient since they have that built in, everything else is handled by GPT-4o, but I might move everything over to Gemini as their pricing is just unbeatable. (2/2)

vm.tiktok.com/ZNeoLrxg5/

I tried to build this with o3-mini in Cursor, but it did not work very well. Claude 3.5 Sonnet got it right on the first try.

Next up: Working on creating cards from YouTube video transcripts and exploring Anthropic's citation feature to give users more insight into which parts of their content are used to create cards, so they can more easily verify the accuracy of the content. (2/2)

docs.anthropic.com/en/docs/buil...

For the more technical folks: This is based on the Visual PDFs feature in Claude 3.5 Sonnet, which ensures that the AI can accurately understand the content of the PDF, including the text and any visuals in the document. (1/2)

docs.anthropic.com/en/docs/buil...

Give it a try: memorics.ignotum.dev

peerlist.io/johschiessl/...

Agree, I always just give them the Tailwind color options because then I can do nice things like use the 100 version for the background and the 500 version for a icon and I know it will look good without having to do any calculations and stuff.

Yes, all designed by myself. I started in Figma. And the rest is just me playing around with the code until I like the look.