Vivekhaldar

(This blog post is derived from this video response on my YouTube channel.) Dwarkesh Patel has been making the case for continual learning for a while, both with guests on his podcast and, most recently, in a blog post and an accompanying video.I agree wi

(This blog post is derived from this video on my YouTube channel, along with a short follow-up about the incentives involved.) I've been writing about specialized agent harnesses and compiling them from task specifications for a while, and I realize the i

I've grown to strongly prefer agents running locally on my own machine. I was doing so even before the latest crop of features from Anthropic and OpenAI that made it possible to remote-control local agents from their mobile apps - something that has made

I used to be a heavy Emacs user. Note the past tense.Over the last few months my Emacs usage has fallen off a cliff, and I want to talk about why, because there's a general lesson in it. My Emacs Commits FlatlinedLike every other Emacs geek, I used to spe

Back when my kids were little and I was working at Google I was dogfooding an internal Maps or Google Now (remember that?) experiment on my Android phone that nudged you to get going for an event that it had inferred you needed to get to-even if it was no

This single command-line pipeline takes an API, stands up an MCP server for it, reshapes its tools, and generates a full interactive UI - with zero hand-written code:mcpknife boot --prompt "Free Dictionary API https://dictionaryapi.dev/" mcpknife mod --pr

Over the last few months, a number of abstractions have shipped from model vendors that help you build systems on top of LLMs-specifically, agents that automate end-to-end workflows. Individually, each abstraction solves a specific problem. But what's com

I've been described as an AI-hype-bro. It's probably fair. The month ChatGPT came out I was mesmerized by it, and it was all I could talk about over lunch with colleagues. I hard-pivoted my career to working on the LLM app layer. I was fortunate enough to

Claude Code has been shipping at a blistering pace. But buried in the recent 2.1.x releases is a fundamental shift in how the tool thinks about extensibility. The changes to skills, slash commands, and subagents might look incremental, but they're pointin

It's not often that we see the birth of a new platform. We're seeing one right now, with an app store for the AI age being built from scratch right before our eyes. OpenAI recently announced that developers can now submit apps to ChatGPT.When that happens