Last weekend, the tech world was buzzing about Manus.im, a Chinese startup that launched a “general AI agent” capable of performing complex tasks. Their impressive demos showed the agent researching topics, creating reports, and completing various tasks – all with minimal human input.
But here’s the thing: after looking under the hood, we realized we could build something remarkably similar using our WordPress plugin, AI Workflow Automation.
What is an ai agent?
An AI agent is a software program that uses artificial intelligence to perform tasks or make decisions based on data inputs. Think of it as a virtual assistant that can automate processes, analyze information, and interact with users. AI agents can handle a variety of functions, such as responding to customer inquiries, generating content, processing data, and more, depending on how they’re programmed.
In the context of the AI Workflow Automation plugin, an AI agent can be created by setting up workflows that define how the agent should respond to different inputs and what actions it should take. This allows website owners to streamline operations and enhance user interactions on their sites.
What is Manus.im (General AI agent)Ā and Why Is Everyone Talking About It?
Manus describes itself as “a general AI agent that bridges minds and actions: it doesn’t just think, it delivers results.” The company claims their agent can handle various tasks for work and life, completing them while you rest.
After testing it out, we confirmed what many suspected – it’s essentially a chain of AI models (Claude Sonnet and Qwen) with function calling capabilities. It’s impressive, but not something that can’t be replicated with the right tools.
Building Our Own General AI agent Version: Meet “M-anus”
As an experiment, we decided to build our own general AI agent using our AI Workflow Automation plugin for WordPress. We affectionately named it “M-anus” (a playful riff on the original).
The setup process was surprisingly straightforward:
- We created a new workflow in our plugin
- Added a Chatbot node as the main interface
- Defined action capabilities for the bot to perform
- Connected research capabilities using the Perplexity API
- Used Claude 3.7 Sonnet for content generation
- Added GPT-4o for title generation and refinement
The Workflow Structure
Here’s what our workflow looks like in the builder:
The workflow follows a simple pattern:
- The user interacts with a chatbot interface
- When specific actions are requested, the bot triggers the appropriate workflow
- The workflow collects information, conducts research, and delivers results back to the user
Testing M-anus Against General AI Agent Of Manus.im
To see how our creation measured up, we took prompts directly from Manus.im’s official examples and ran them through our system.
For instance, we asked both systems to research “the largest real estate brokerages in the top 10 major metro areas.”
General AI agent Manus.im Result:
Our M-anus Result:
As you can see, both systems delivered comprehensive reports with similar information. In some ways, our system provided more detailed regional breakdowns.
How It Works Behind the Scenes
When a user sends a message to M-anus, here’s what happens:
- The Chatbot node analyzes the request to determine intent
- If research is needed, it triggers the Research node with Perplexity API
- The results are sent to Claude 3.7 Sonnet to structure and expand the content
- GPT-4o helps refine titles and headings
- A new post is generated using the Post Node automatically
- The results are delivered back to the user
We can even track the cost and performance of each component:
The entire process for our real estate example cost just $0.0762 in API calls – remarkably efficient for the value delivered.
Why WordPress Makes This Even Better
The key advantage of building this in WordPress with our plugin is that we don’t need to waste tokens generating UI elements. WordPress already provides:
- A ready-made front-end interface
- User authentication
- Content storage and management
- The ability to publish results as posts or pages
- Email capabilities built-in
Build Your Own General AI Agent
The exciting part is that anyone using our AI Workflow Automation plugin can build their own general AI agent. No coding required – just drag and drop nodes in our visual workflow builder.
Here’s what you’ll need:
- WordPress site
- Our AI Workflow Automation plugin
- API keys for your preferred AI models (OpenAI, Anthropic, etc.)
- A clear idea of what tasks you want your agent to perform
The Chatbot node with its Actions feature is particularly powerful. You can define specific actions your agent can perform, such as:
- Researching topics
- Creating content
- Sending emails
- Posting to social media
- Analyzing data
- And much more
You can download M-anus as a template or a starting point and build on that:
- Download the workflow from this link
- Download and install the Pro version of AI Workflow Automation
- Open a new workflow, and on the bottom left click on Import/export workflow
- Drag and drop the workflow you just downloaded to that box
- Save the workflow and start playing with M-anus!
Final Thoughts
While companies like Manus.im are making headlines with their general AI agents, the technology to build similar systems is increasingly accessible. Our experiment shows that with the right tools, anyone can create powerful AI agents without extensive coding or infrastructure.
AI Workflow Automation brings this capability to WordPress, allowing businesses and creators to build custom AI agents tailored to their specific needs.
Have you tried building your own AI agent? Let us know in the comments!
Want to build your own general AI agent like M-anus? Check out our AI Workflow Automation plugin and start creating today!
