LanMate Overview
LanMate is a local-first AI work execution platform by Lanxin — known in Chinese as 蓝莓办公 (Lanmei Bangong, “Berry Office”). It goes beyond chat — it gets the job done on your local machine: reading and writing files, running code, generating documents, and controlling browsers, with every change producing a reviewable, undoable diff. Once your organization enables it, tasks can also run in a private cloud sandbox.
Core Capabilities
| Capability | Description |
|---|---|
| Local Execution | Files, code, and data processed entirely on your machine; built-in offline Python environment |
| MateBot Personal Assistant | A sidebar assistant that logs items, watches deadlines, collects messages, dispatches work, and brings receipts back — nothing falls through the cracks (see MateBot) |
| Cloud Execution | Once enabled by your organization, tasks can run in a private cloud sandbox — a second run location alongside local (see Cloud) |
| Office Documents | Built-in Word / Excel / PPT / PDF editors, with 18 official document templates and 16 PPT templates out of the box |
| Lansenger Collaboration | Sign in with Lansenger QR code and bind a personal bot; send commands from your phone, your computer executes, and approval cards reach Lansenger directly |
| Multi-Agent Orchestration | Taskboard dispatch, expert rooms, scheduled jobs — multiple agents working in parallel |
| Organization Control | Three-tier policies for models and skills, server-side API key hosting, full audit trail |
Two Surfaces
- Task sessions: one conversation per task, with the tool surface chosen for the work — files, code, documents, and browser all in one place
- MateBot: the persistent personal assistant. She does not compete with execution sessions; she records, watches progress, dispatches, and accepts
The board is the project’s execution surface (scheduling, review, branches, worktrees); MateBot is the personal surface (someone owns it, a receipt comes back, someone accepts it). The layers do not replace each other.
Platform Support
| Platform | Status |
|---|---|
| macOS Apple Silicon (arm64) | Fully supported |
| Windows x64 | Fully supported |
| Linux (Kylin / UnionTech) | Supported (ARM and x86) |
| HarmonyOS PC | Supported (chart generation, diagrams, image cutout, and image/PDF text recognition excluded) |
Login Options
- Lansenger QR code sign-in: Scan with the Lansenger app to sync org policies and available models.
- Anonymous mode: Use all local features (conversations, files, code, skills) without signing in—handy for personal trials.
Workflow at a Glance
One complete loop of use:
- New task: pick a project and run location, describe the need in one sentence — or hand it to MateBot first
- Provide context: @ reference files, sessions, or tasks; drag attachments into the composer
- Agent executes: tool calls are visible in real time; approval cards pop when confirmation is needed
- Review output: diffs, document previews, and files in the right panel
- Iterate or deliver: keep refining, or export, commit, and archive
Data is processed locally by default; with cloud execution or organization models enabled, data flows per your organization’s policies. Approval and the sandbox control the boundaries, and artifacts are reviewable and reversible.