Recently,OpenClaw has been continuously trending. In the past, when we used AI, it was just an ordinary chatbot, but OpenClaw turned AI into a 24/7 AI employee that never slacks off.
Imagine this: while we sleep, AI collects information for us in the background; while we slack off, AI keeps working; we can even just send a message on our phone, and AI will respond at any time.
The AI community has directly whipped up a "raising lobsters" storm.
But after actually raising 3 crayfish, I found that OpenClaw actually has an extremely high barrier to entry. All kinds of environment configurations and error codes directly scare off non-technical beginners. After finally getting it deployed and running a task, one look at the Token consumption really feels like the sky is falling...
Late last night, Zhipu suddenly launched a domestic version of the little lobster——AutoClaw(Australian lobster).
After testing overnight, I found that AutoClaw has almost solved all our pain points in "raising lobsters."
A quick summary of this "Australian lobster's" killer features:
- The full-blooded version of OpenClaw for local use, one-click deployment, ready to play right after installation;
- Provides free models and high-cost-performance packages, greatly reducing Token consumption costs;
- No network restrictions, one-click access to Feishu, seamlessly calling various lobster Skills.
- Freely switch models, supporting GLM, DeepSeek, Kimi, etc.;
- Independently set up the workspace to ensure data security.
This lobster is truly amazing. This article will provide a hands-on walkthrough of how to raise an "Australian lobster" with zero barriers.
01. The first domestic lobster client
AutoClaw directly takes care of the environment configuration of the original OpenClaw that has discouraged countless people. We open the official website and download the corresponding installation package (both Windows/Mac are supported).
Official website：https://autoglm.zhipuai.cn/autoclaw
Foolproof deployment
Double-click to install, open the client, and log in with mobile phone number verification.
At this step, we have completed the local deployment of AutoClaw~ We can start using it directly~ It really is so easy anyone can do it.
AutoClaw has a built-in new model Pony-Alpha-2 specially created for lobsters, and directly gives free quota~ You can start playing at zero cost.
Click the quick configuration button on the page to let AutoClaw initially get to know you.
It is recommended to turn off restricting the file access scope; otherwise, AutoClaw cannot read files outside the working directory, which will make it very uncomfortable to use. For example, if we want AutoClaw to process a document on the desktop, AutoClaw will report an error due to insufficient permissions.
After completing the configuration, we can start chatting with AutoClaw.
From opening the official website to sending the first command, the whole process took less than a minute~ A zero-barrier, localized OpenClaw that can also be experienced for free is really incredible.
Model freedom
When it comes to raising lobsters, what friends care about most is Token consumption. After all, every string of characters is real money.
But in AutoClaw, the perspective is completely opened up. AutoClaw not only provides free model quota, but also supports access to any model you want, such as DeepSeek, Kimi, OpenAI, GLM, and so on.
It also provides GLM model packages of different tiers, making it convenient for everyone to choose according to their needs.
The best part is its backend panel. Token and credit consumption are clearly visible, so you don't have to worry about going "bankrupt" while using it.
Built-in Skills
Although deployment is simple, AutoClaw is not vague at all when it comes to getting work done.
AutoClaw comes with some very useful OpenClaw Skills built in, which can be called directly.
Prompt: I want to do an in-depth research project on: [Sustainable Business Models for One-Person Companies]. Please help me: 1. Break down this question from multiple angles; 2. Search Chinese and English sources and cross-verify the data; 3. Provide a deeply researched report with a clear structure, data support, and independent insights;
AutoClaw will automatically break down the task and call browser tools to search, read, and organize across the web step by step.。
It also captured the latest information from March 5, and both timeliness and accuracy are quite good:
Tutorial for integrating with Feishu
Only by integrating AutoClaw with IM can you truly achieve automated office work.
In Settings, find IM Channels, add a Feishu channel, and enter the APP ID and App Secret. (Feishu currently supports 1 million calls per month for individual developers. Ordinary people cannot possibly use that many!)
For friends with zero experience, when creating a Feishu bot for the first time, you can refer to the configuration tutorial here.https://zhipu-ai.feishu.cn/wiki/TMNewihmoiyQ2jkPhpDccgOunAe。
After configuration is complete, we will have an AI bot in Feishu. In the group settings, we can add the AI bot to the group chat,and then through Feishu we can have AutoClaw process messages, automatically execute tasks, call Skills... very convenient.。
Let's try @Tangyuan in the group:
Prompt: Collect inspiration from public materials, extract themes and character relationships, generate an original short story framework, and create a short story.
After AutoClaw receives the group message, it will immediately reply with a little keyboard-typing emoji, meaning "I saw it, I'm working on it"~
After a while, AutoClaw sent me the novel it generated.
02. Some sharing
After going through a round of experience, AutoClaw really brings the five major advantages of localization, zero threshold, Feishu multi-channel integration, Skill expansion, and model freedom to the extreme. This domestic lobster is truly great.
If you also want to unlock an AI employee at low cost, I highly recommend AutoClaw plus GLM's Coding Plan package. It's cheap and easy to use! You won't be so anxious about Token consumption that you don't dare to try it~
But GLM's Coding Plan is very hard to get. I suggest setting an alarm in advance and going to wait at 10 a.m. sharp.
In recent years, AI has been evolving faster and faster, with new models, new products, and new ways to use them every day. The white-collar class, which was once thought to be unaffected, is also facing a major industry reshuffle with the spread of automation tools like Auto.
In the future, the white-collar class will polarize, leaving only two roles:
One is being able to manage multiple AIs to solve complex problems; the other is serving AI and becoming part of computing power, algorithms, and data.
Original link:Zhipu launches AutoClaw, install the full-power version of OpenClaw in one minute