AI

  • Everyone is abuzz about Deepseek after NVidia and other stocks had a torrid Monday (and then a rebound today), but if you only read one piece Ben Thomspson’s Deepseek FAQ. Two, well then take a look at the technical paper - DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning. Seperately the R1 model is making its way onto providers fast with availability on Cloudflare already.

  • Agent Experience is a new concept that Matt Billman, CEO of Netlify, has just introduced. I love this comment “Agents will far more frequently be collaborators and extensions of humans, rather than replacements. Both will wildly increase the productivity and the ability of a single human being.” That feels pretty on-point for at least the software world for now.

  • Vercel continue to build out AI SDK, their unified API for building agents with LLMs, this time with the addition of Googles Gemini 2.0 Flash Experimental model. Lets be honest, being the ower of npm i ai is a pretty slick developer marketing move on Vercels part.

General Interest

Data