This is a new paradigm for interacting with Claude that is significantly more "inline" with all the other human activity org-wide. Once you do all of ...
Andrej Karpathy describes a new paradigm for interacting with Claude where the LLM integrates into an organization as a persistent, asynchronous team member with access to shared tools and context. He frames this as the third major redesign of LLM UXUI, evolving from a website to an app to now a self-contained entity working alongside human teams.
Background
Andrej Karpathy (a prominent AI research scientist, formerly at Tesla AI and OpenAI) is reacting to Anthropic's "Claude Tag" feature, which lets users bring the Claude AI assistant directly into Slack as a collaborative team member that can be @-mentioned and assigned tasks. Karpathy argues this represents a third major phase in how we interact with large language models: first as websites (like ChatGPT), second as downloadable desktop apps, and now as persistent, asynchronous "team members" embedded within existing enterprise communication and tooling infrastructure. The key distinction is that Claude Tag requires significant backend engineering (tool integrations, memory, security, compute environments) to make the AI feel like a seamless co-worker rather than a separate application you context-switch to.