Kimi Work: The 24/7 AI Desktop Agent for Knowledge Workers

Kimi Work

37ms7892💬 9
Kimi Work: The 24/7 AI Desktop Agent for Knowledge Workers

I introduce Kimi Work, a local AI agent designed to automate your entire workflow. It runs 24/7 with a built-in Cron engine, browses the web autonomously via WebBridge, and coordinates Agent Swarms to solve complex tasks. With native support for global market data and Python execution, it acts as a system-level digital employee that handles repetitive tasks while you sleep.

"You provide the prompt, Kimi handles the clicks, scrolls, and research."

HN discussion

  • Users express significant wariness about granting Kimi Work direct access to their desktop and filesystem, contrasting this risk with the relative safety of using APIs via OpenRouter.
  • Practitioners argue that switching between AI agents is frictionless, noting that tools like Claude Code can be reconfigured to use the K3 model in under 30 seconds.
  • Critics contend that the 'Chinese gotcha' narrative regarding censorship is often overstated compared to the arbitrary refusal behaviors observed in models like Fable and Gemini.
  • Security-conscious users advocate for open-source agent harnesses to ensure transparency, citing the Grok fiasco as evidence of the risks inherent in closed-source agentic systems.
  • Developers highlight that building custom AI agents is accessible by looping LLM calls with basic tools, allowing users to create tailored solutions without relying on proprietary interfaces.

More from this day · 2026-07-20