Kimi AI Introduces Kimi Work: Always-On Desktop Agent With Web Automation, Scheduling Engine And Swarm-Based Architecture

Kimi AI, a project developed by Beijing-based startup Moonshot AI, has announced Kimi Work, a local desktop AI agent designed to automate tasks across files, web browsers and code execution environments. The company says the product is intended to support deeper, more complex workflows by combining local folder access, autonomous web navigation, background Python execution and scheduled task handling in a single system-level tool.

According to the announcement, Kimi Work is built to operate continuously rather than as a one-off assistant. A built-in scheduling engine allows tasks to run on a daily, hourly or conditional basis, enabling routine processes such as drafting briefings, processing datasets or launching scripts at preset times. The company also said users can keep the computer awake overnight to ensure tasks continue without interruption.

A central feature of the platform is WebBridge, the browser automation layer that enables the agent to move through websites, open tabs, gather information and complete multi-step online actions with minimal user input. The agent can be instructed to monitor official sources, collect market data or compile material into local files, while handling the mechanics of browsing, clicking and typing on its own.

Swarm Architecture, Financial Data Integration, and Persistent Memory

The announcement also points to a swarm-based architecture that allows multiple specialized agents to work in parallel on different parts of the same task. Moonshot AI says this approach is meant to accelerate research and execution, particularly when a project requires sorting through several layers of information before producing a final output. Once the work is completed, the system can convert findings into documents, presentations or spreadsheets.

Finance appears to be one of the main use cases highlighted in the rollout. Kimi Work is said to include native access to market data covering A-shares, Hong Kong stocks and US equities, reducing the need for custom API setups. The company says users can retrieve earnings information, review market trends and reconcile spreadsheets through natural language interaction.

Other features announced for the desktop product include a memory system that retains preferences, prior decisions and contextual notes over time. According to Moonshot AI, this is intended to make the agent more responsive to recurring workflows and long-term usage. The launch positions Kimi Work as part of a broader effort to bring agentic AI tools closer to everyday desktop productivity and professional analysis.

Kimi Work is available for both macOS (Apple Silicon) and Windows, ensuring compatibility across the two major desktop operating systems and enabling users to deploy the agent in standard professional computing environments.

The post Kimi AI Introduces Kimi Work: Always-On Desktop Agent With Web Automation, Scheduling Engine And Swarm-Based Architecture appeared first on Metaverse Post.