Skip to content

Latest commit

 

History

History
217 lines (151 loc) · 8.24 KB

collaboration-models.md

File metadata and controls

217 lines (151 loc) · 8.24 KB
sidebar_position
4

Collaboration Models

This document introduces the four collaboration models in Wegent, helping you choose the most suitable pattern for building your agent teams.

Collaboration Models Overview

Collaboration models define how multiple Bots interact and work together within a Team. Choosing the right collaboration model can significantly improve team efficiency and task completion quality.

Four Collaboration Models

Model Icon Characteristics Use Cases
Pipeline 🔄 Sequential execution, linear flow Code development workflow, content production pipeline
Route 🎯 Intelligent routing, on-demand assignment Customer service, problem classification
Coordinate 👥 Parallel coordination, result aggregation Multi-perspective analysis, parallel task processing
Collaborate 💬 Free collaboration, shared context Brainstorming, complex problem solving

Selection Decision Tree

Start Model Selection
    │
    ├── Does task have fixed order?
    │   └── Yes → Pipeline (Sequential Model)
    │
    ├── Need content-based task routing?
    │   └── Yes → Route (Routing Model)
    │
    ├── Can tasks be parallelized?
    │   └── Yes → Coordinate (Coordination Model)
    │
    └── Need free discussion?
        └── Yes → Collaborate (Collaboration Model)

🔄 Pipeline

Overview

Pipeline is the simplest and most intuitive collaboration model. Bots execute sequentially in a predefined order, with each Bot's output becoming the input for the next, forming a complete processing pipeline.

Workflow

Task Input → Bot 1 → Bot 2 → Bot 3 → Bot 4 → Final Result

Core Features

  • ✅ Sequential Execution: Strictly follows defined order
  • ✅ Data Passing: Previous Bot's output automatically passes to next
  • ✅ Simple and Clear: Easy to understand and maintain
  • ✅ Error Isolation: Failures can be quickly localized

Use Cases

  • Software Development Workflow: Requirements Analysis → Design → Implementation → Code Review → Testing
  • Content Production Pipeline: Content Creation → Editing → Formatting → Quality Review
  • Data Processing Pipeline: Data Collection → Data Cleaning → Data Transformation → Data Analysis

🎯 Route

Overview

Route model uses a Leader Bot to intelligently route tasks to the most suitable expert Bot based on task content. Ideal for scenarios requiring different handling paths based on varying conditions.

Workflow

Task Input → Leader Bot (Routing Decision)
              ├── Frontend Issue → Frontend Bot
              ├── Backend Issue → Backend Bot
              ├── Database Issue → Database Bot
              └── DevOps Issue → DevOps Bot
                      ↓
                   Result

Core Features

  • ✅ Intelligent Routing: Leader Bot analyzes task and selects best expert
  • ✅ Specialized Roles: Each Bot focuses on specific domain
  • ✅ Flexible and Efficient: Dynamically selects execution path based on needs
  • ✅ Extensible: Easy to add new expert Bots

Use Cases

  • Intelligent Customer Service: Customer Question → Classification → Technical Support/After-Sales/Sales
  • Technical Support Platform: Technical Issue → Analysis → Frontend Expert/Backend Expert/Database Expert
  • Content Distribution System: Content Submission → Classification → News/Tech/Entertainment Channel

👥 Coordinate

Overview

Coordinate model has a Leader Bot decompose tasks and assign them to multiple Bots for parallel processing, then collect and aggregate all Bot results to form the final comprehensive output.

Workflow

Task Input → Leader Bot (Task Decomposition)
              ├── Analyst Bot ──┐
              ├── Data Bot ─────┼→ Leader Bot (Result Aggregation) → Comprehensive Result
              └── Research Bot ─┘

Core Features

  • ✅ Parallel Processing: Multiple Bots work simultaneously for efficiency
  • ✅ Task Decomposition: Leader breaks down complex tasks
  • ✅ Result Aggregation: Leader integrates multiple perspective outputs
  • ✅ Comprehensive and In-depth: Analyzes problems from multiple angles

Use Cases

  • Market Research Analysis: Market Research Task → Competitor Analysis/User Research/Trend Analysis → Comprehensive Report
  • Code Review: Code Review Task → Security Review/Performance Review/Code Quality → Review Report
  • Content Creation: Article Topic → Material Collection/Case Study/Data Analysis → Complete Article

💬 Collaborate

Overview

Collaborate model allows all Bots to freely discuss and collaborate in a shared context, without fixed execution order or clear role divisions. All Bots can see each other's messages and respond based on the overall conversation.

Workflow

Task Input → Shared Context Space
              ↕        ↕        ↕        ↕
           Bot 1    Bot 2    Bot 3    Bot 4
              └────────┴────────┴────────┘
                         ↓
                Collaborative Output

Core Features

  • ✅ Shared Context: All Bots see complete conversation history
  • ✅ Free Discussion: No fixed order, Bots can participate anytime
  • ✅ Collective Intelligence: Promotes creativity and multi-perspective thinking
  • ✅ Dynamic Interaction: Bots can respond to others' viewpoints

Use Cases

  • Brainstorming Sessions: Creative Topic → Product Manager/Designer/Engineer/Marketing → Creative Solutions
  • Technical Solution Design: Technical Challenge → Architect/Developer/DBA/DevOps → Technical Solution
  • Complex Problem Diagnosis: System Issue → Backend/Frontend/DBA/DevOps → Root Cause

Model Comparison

Feature Comparison Table

Feature Pipeline Route Coordinate Collaborate
Execution Order Fixed sequence Dynamic single path Parallel execution No fixed order
Bot Interaction Unidirectional passing Leader → Expert Leader coordination Full interaction
Context Sharing Partial (sequential) Independent context Leader aggregation Fully shared
Team Size 3-6 Bots 1 Leader + N experts 1 Leader + 3-5 experts 3-8 Bots
Execution Efficiency Medium (serial) High (single path) High (parallel) Medium (discussion)
Ease of Use ⭐⭐⭐⭐⭐ ⭐⭐⭐⭐ ⭐⭐⭐ ⭐⭐⭐
Flexibility ⭐⭐ ⭐⭐⭐⭐ ⭐⭐⭐⭐ ⭐⭐⭐⭐⭐

Selection Guide

  • Standardized Process → Pipeline
  • Need Specialized Division (Single Expert) → Route
  • Need Specialized Division (Multiple Experts) → Coordinate
  • Need Free Discussion → Collaborate

Best Practices

General Recommendations

  1. Start Simple: Prefer Pipeline or Route first, avoid over-engineering
  2. Control Team Size: Each model has recommended Bot count, too many reduces efficiency
  3. Clear Role Definition: Each Bot's responsibilities should be clear and specific
  4. Design Good Prompts: Good prompt structure includes role definition, responsibilities, input/output requirements

Key Points by Model

Model Key Points
Pipeline Control step count (≤6), single responsibility per step
Route Clear routing rules, include default handling path
Coordinate Leader needs strong aggregation capability, control parallel count (3-5)
Collaborate Define clear goals, set discussion termination conditions

Related Documentation

Choosing the right collaboration model is key to success! 🚀