They live on your computer. Assign work to one, have several team up, or send instructions from your phone — things keep moving around the clock.
A minute, start to finish. Native installers for Windows 10+ and macOS 12+, including Apple silicon and ARM64.
Claude Code, Codex and Gemini CLI are picked up if you have them. Nothing installed? 20+ built-in assistants work out of the box.
Plain language is enough. Decks, spreadsheets, code, retouched photos — the output lands in your own folders, ready to edit.
The chat window is where you'll spend most of your time in YggLab. It isn't a box you type into and then wait — while a task is running you can interject, queue up what's next, dial back permissions, and see exactly what it's doing at every step.
While the main task keeps running, you can ask a quick question about the current context. The answer comes back on its own, and the main task never notices — no waiting for it to finish, no second chat window.
While the AI is still answering, queue up the next few steps. Queued instructions can be edited, reordered, or dropped — you don't have to sit there feeding them in one at a time.
Three tiers: plan only, auto-accept edits, fully autonomous. Any single tool call can also be approved just this once or allowed from now on. How much rope it gets is your call.
What command ran, in which directory, the exit code, how long it took, which files changed, lines added and removed — every tool call is laid out as a card, so when something goes wrong you can see which step did it.
YggLab finds the Claude Code, Codex, Gemini CLI and 20+ other CLIs already on your machine. Run them in parallel, give each its own job, or have them team up — all in one window.
|
|
|
＋
20+ assistants live on your computer, each with its own specialty. Nothing to configure — open YggLab and start assigning work.
You don't configure them one by one. Pick by specialty — someone to write the docs, someone to do the visuals — from Documents, Design & visualization, Business and Development, and the engine matches an agent to each. Pick two who overlap and it'll tell you the extra member only adds coordination cost. Pin the teams you use often; swap a member or its engine whenever it isn't working out.
W
D
S
D
Web
＋
Seven remote channels: Slack, Discord, Telegram, WeChat, WeCom, Feishu and DingTalk — plus a built-in WebUI you sign into by scanning a QR code. Send a message on any of them to assign work, or drop in a PDF, Word file or image; the result comes back to the same chat. You're out; the agent keeps going on your machine.
Schedule agents to run on their own. Code review every morning, backups every night, a report every Monday. You sleep, they work.
Flexible model configuration, extensible skills, one place for MCP — so the agents actually work the way you want.
No API key required.Sign in to a YggLab account and cloud models are configured for you — install and start working. Prefer your own keys? 28 model platforms are one click away, and you can point at any OpenAI- or Anthropic-compatible endpoint. Different agents can use different models, connecting straight to the API with nothing of ours in between.
Browse and install community skills to extend your assistants. You can build and share your own, too.
Add an MCP server once and every agent picks it up. No per-agent setup — do it once, they're all connected.
PowerPoint, Excel, Word and PDF, generated natively. Nothing uploaded, no third party — the files stay on your computer.
AI rarely nails it first try; what actually decides your speed is how easily you can fix what it got wrong. The output panel isn't read-only — decks, spreadsheets, documents, images, HTML and PDFs can all be edited in place, diffed, and rolled back.
Whatever the AI got wrong, fix it in the preview panel and save — no opening Word, Excel or an editor, no copying back and forth.
Which files it touched, how many lines added and removed — all listed. Discard the changes you don't want one by one instead of throwing the whole thing out.
Save a snapshot at any point and roll back if something breaks. You don't have to withhold permissions just in case.
Watch the document fill in as it's written; newly generated office files open in the preview panel on their own.
See the page as it's being built, and click any element in the preview to jump to the code behind it.
Compare several files, or split the view; or hand one off to Word or Keynote in a click. PDFs just open, and Mermaid flowcharts and sequence diagrams render as diagrams instead of a wall of code.
Conversations, files and task history are stored locally. Models are reached through their official APIs with no relay server in between. Everything runs on your machine, and everything it does is auditable.
What people ask most before installing.
Want to see it actually working? Read the write-ups. New releases and new ways to use it show up there first.
In China you can also follow us on WeChat — scan the code; the account shows up as 猿滚滚.
macOS · Apple silicon
Windows · x64