Moonshot AI has introduced Kimi Work, a desktop AI agent that runs locally on macOS and Windows, breaking from the cloud-based paradigm that has defined most agent tools over the past two years. Announced this week with downloads available for both platforms, Kimi Work reads local files, drives your real browser, and executes scheduled tasks directly on your machine. It targets knowledge workers whose productivity bottleneck is access to live files and authenticated sessions — precisely the assets that cloud agents cannot reach.
What Is Kimi Work? The Desktop Agent Built on Kimi K2.6
Kimi Work is a downloadable application, not a web chat. You give it goals in plain language, and it acts on your local machine. It runs on Kimi K2.6, Moonshot’s flagship model — an open-weight Mixture-of-Experts architecture released on April 20, 2026. K2.6 activates approximately 32 billion parameters per token and carries a 256K-token context window, enabling long, multi-step workflows without losing state.
This local execution model is a deliberate departure from the agent tools that dominated the past two years. Most of those run in the cloud: you type a goal, a remote server spins up a sandbox, and a hosted browser executes the actions. Kimi Work instead reaches the files, browser sessions, and applications you already use every day.
Four Building Blocks That Define Kimi Work
Four technical components give Kimi Work its capabilities. Understanding them clarifies what the product can and cannot do.
Agent Swarm: Kimi Work can run many sub-agents in parallel on your machine. The swarm scales to 300 sub-agents, splitting a task into parts and coordinating the results. The K2.6 swarm is documented up to 4,000 coordinated steps, meaning complex, multi-stage workflows are feasible.
WebBridge: This browser extension lets the agent control your browser like a human operator. It searches, scrolls, extracts data, and fills forms across tabs. Because WebBridge uses your real, logged-in session, it inherits your existing cookies and authentication — no need to re-enter credentials or rely on a sandboxed environment.
Cron Scheduling Engine: A built-in scheduler runs jobs on a daily, hourly, or conditional basis. Triggers include LLM agent calls, Python scripts, and shell scripts. A “Keep Computer Awake” toggle prevents overnight jobs from stalling due to system sleep.
Local Files and Code Execution: The agent reads folders you mount and runs Python in the background. Original files remain untouched unless you explicitly approve a change, giving you control over your data at all times.
The desktop app also ships with pre-integrated financial data for A-shares, Hong Kong stocks, and US equities. This removes the need for custom API configuration. Finished research can be converted into PowerPoint decks or Excel sheets directly from the agent.
Practical Use Cases for Knowledge Workers
Kimi Work addresses real workflow bottlenecks that cloud agents cannot solve because they lack access to local files and authenticated sessions. Four scenarios illustrate the difference.
Document triage: Point the agent at a folder of quarterly PDFs and ask it to summarize them into one document while keeping originals intact. The swarm assigns one reader per file, then merges the findings into a single output.
Web data collection: Tell WebBridge to pull historical prices for three tickers. It opens your browser, sets the date range, and extracts the tables. Python then normalizes the columns and writes an Excel workbook — all on your machine.
Scheduled briefings: Define a 7:00 AM job in the cron engine. Each morning, the agent gathers headlines and drafts a markdown briefing. With the “Keep Computer Awake” toggle enabled, the job runs reliably overnight.
Office generation: Ask for a short market-brief deck after a research pass. The agent drafts sections in parallel and renders native slides without requiring any external API or upload.
Kimi Work vs. Cloud-Based Agent Platforms
The core difference between Kimi Work and a typical cloud agent is where execution happens and what resources the agent can access.
| Dimension | Kimi Work (local) | Typical cloud agent |
|---|---|---|
| Execution location | Your desktop | Vendor servers |
| File access | Mounts your local folders | Uploaded or sandboxed files |
| Browser | Your real, logged-in browser via WebBridge | Hosted virtual browser |
| Scheduling | Built-in cron engine | Often external or limited |
| Underlying model | Kimi K2.6 | Vendor’s hosted model |
| Setup | Install app, grant folder access | Zero-install, open a tab |
| Security responsibility | Falls on the user | Falls on the vendor |
Neither approach wins outright. Local execution keeps sensitive data on your device and reaches real files and authenticated sessions. Cloud execution trades that control for zero-setup convenience and managed safety. Your choice depends on whether your workflows require access to local resources or can operate entirely within a sandboxed environment.
Scheduling Tasks with the Built-in Cron Engine
Kimi Work uses standard cron syntax for scheduling, not a proprietary API. The five fields — minute, hour, day-of-month, month, and day-of-week — follow the same convention used by Linux system administrators for decades.
A daily briefing job is configured by pairing a cron schedule with a plain-language task description. For example:
Schedule: 0 7 * * *codecodecodecode (every day at 07:00)
Task: “Draft today’s market briefing and save it to ~/KimiWorkspace/briefing.md. Ask before writing.”
The approval gate applies to every file write and web action by default. Users can disable this gate — referred to as YOLO mode — but the default setting prompts before any change is made to your local files or browser state.
Who Should Use Kimi Work Now
Kimi Work is best suited for professionals who regularly work with local files, authenticated web sessions, and recurring data tasks — analysts, researchers, marketers, and operations teams who need an agent that reaches the tools they already use. It is less suited for users who prefer zero-install setups or who want a vendor to manage all security boundaries. For those whose work depends on accessing live documents and real browser sessions, Kimi Work offers a local alternative that cloud agents cannot match. The application is available now for macOS (Apple silicon) and Windows, making it immediately accessible to anyone who wants to try running an agent swarm on their own desktop.
/i.s3.glbimg.com/v1/AUTH_08fbf48bc0524877943fe86e43087e7a/internal_photos/bs/2026/q/p/xkrnfASUWa5SSkay0x6A/marvels-wolverine-20260907193401.jpg "Marvel’s Wolverine Best Techniques to Unlock First")
