Prateek Joshi: LinkedIn Content Strategy & Writing Style

Researchers at Google just published a fascinating finding: Simply repeating your prompt twice can dramatically improve LLM performance. In some cases taking accuracy from 21% to 97% on specific searc…

It’s the Wild West in AI! Industrial scale distillation attacks against Claude. Looks like DeepSeek, Moonshot AI, and MiniMax created a large number of accounts and had 16 million exchanges with Claud…

Software engineering is eating all the tokens. Today coding agents, IDE assistants, refactoring loops, test generation, and CI integrations are some of the most token-intensive workflows in the AI e…

OpenAI just published solution attempts for the ten problems from First Proof. People usually wonder if AI can solve mathematics or it just searches for existing literature to spit out solutions that…

Not sure what to make of this. Enormous headcounts! But I suppose Zoom having 12,743 employees makes sense because it takes a lot of effort to accurately time the software update when you’re running l…

Coding is still the highest-intensity use case for foundation models. Claude Code is clearly the undisputed leader in coding LLMs. And Codex is still playing catch up. So how will the OpenClaw move by…