The Kimi Work Beta version has commenced internal testing, marking its debut alongside the newest test iterations of the Mac and Windows clients for Kimi. Specifically crafted for the needs of knowledge professionals, Kimi Work stands as a versatile local Agent, with Kimi Code at its core. It provides not only the essential functionalities of a local Agent but also taps into the wealth of expertise and expansive database of the online Kimi Agent.

Leveraging the integrated Kimi WebBridge solution, users can effortlessly articulate their task objectives in natural language. Kimi Work then takes charge, automatically dissecting tasks, managing them concurrently, invoking the requisite tools, harnessing browser capabilities, creating and organizing folders on the user's computer, and ultimately presenting the completed work results.

Moreover, Kimi Work boasts support for Agent clustering, empowering the autonomous formation of teams comprising up to 300 sub-Agents. Remarkably, the Kimi Work Beta client's coding was significantly influenced by Kimi Code, with engineers utilizing the Coding Agent tool to finalize development within a mere week. Throughout this development phase, an impressive output of over 50,000 lines of effective code was generated, with a staggering 92% autonomously crafted by AI.