Anthropic is coming for EVERYTHING

TL;DR

• 1

  Claude tag is a Slack-integrated AI that ingests all company data: It reads conversations, documents, and has context on every employee. 65% of Anthropic's product team code now comes from their internal version, according to the video.

• 2

  This is the third evolution of LLM UX: Andre Karpathy describes it as moving from "a website you go to" to "an app you download" to now "a self-contained, persistent, asynchronous entity with orgwide tools and context working alongside teams of humans."

• 3

  Anthropic aims to own all knowledge work platform risk: Matthew Berman warns that Claude tag creates "context lockin" rather than model lockin, meaning companies rent back their own business from Anthropic with uncapped token costs that could replace human salaries with unlimited AI consumption.

• 4

  The feature threatens to eliminate software UI entirely: If agents operate SaaS tools on behalf of customers, users no longer log in. The workflows become valueless because AI can write code to replace them, reducing all software to merely its underlying database.

• 5

  Government intervention and open-source models are the only safeguards: Without competition and the ability to own one's own business context, a single company could end up controlling all knowledge work and all software, creating a permanent underclass dependent on token-based productivity.