Skip to main content
Shinkai Web allows multiple AI agents to work together inside the same workflow.   This feature lets you combine different skills — research, writing, analysis, editing, creativity — into a single automated process.

​
What Is Agent Collaboration?

Agent Collaboration means you can choose two or more agents and connect them in a single shared task.   Each agent contributes with its own:
  • personality  
  • instructions  
  • tools  
  • workflow role (researcher, writer, editor, reviewer, creator, etc.)
This creates a chain of reasoning and production that a single agent cannot match.

​
Why Collaboration Matters

With multiple agents, you can:
  • automate multi-step workflows  
  • mix different skill sets  
  • reduce human intervention  
  • improve output quality  
  • create reusable pipelines  
  • scale your productivity in the browser
Shinkai Web makes collaboration accessible to everyone — no coding needed.

​
How to Start a Multi-Agent Workflow

  1. Open Agents from the sidebar  
  2. Select create agent or choose workflow presets (if available)  
  3. Choose to start from scratch or choose a template.
  4. Fill in the main fields:
   - Name      - Description      - Instructions
  1. Choose which tools the agent can use.
  2. Choose which agents you want to collaborate (2, 3 or more)
  3. Start the workflow  

​
Defining Agent Roles

When setting up collaboration, you can define:
  • Primary agent  
  Leads the workflow and produces the final answer.
  • Supporting agents  
  Provide info, analysis, content or structure.

​
Collaboration Controls

During a multi-agent workflow you can:
  • pause the workflow  
  • skip an agent  
  • regenerate only one agent’s response  
  • modify instructions mid-flow  
  • add or remove agents  
  • restart the workflow with the same setup  
You maintain full control from the browser.

​
Tool Usage in Collaboration

Agents in a workflow can use their own tools independently:
  • Research agent → web search  
  • Creative agent → image generation  
  • Video agent → text-to-video  
  • Data agent → document parsing  
This makes workflows extremely powerful.

​
Tips for Great Collaboration Results

​
Keep instructions clear  

Tell the agents exactly what you want each role to do.

​
Give context upfront  

Provide files, links, data, or examples.

​
Use specialized agents  

Writer + Researcher + Editor always beats a “generalist.”

​
Iterate  

You can rerun the pipeline until the output matches your expectations.

​
When to Use Agent Collaboration

Use collaboration for:
  • content creation pipelines  
  • research workflows  
  • document analysis + summarization  
  • report generation  
  • marketing assets  
  • long-form content  
  • code review / documentation  
  • any process requiring multiple expert “roles”
Next Steps Now explore the remaining advanced features:
  • x402 Payments on Web-Use credits to run advanced agents and higher-volume tasks.