Skip to main content

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​

ModelIconCharacteristicsUse Cases
PipelineπŸ”„Sequential execution, linear flowCode development workflow, content production pipeline
Route🎯Intelligent routing, on-demand assignmentCustomer service, problem classification
CoordinateπŸ‘₯Parallel coordination, result aggregationMulti-perspective analysis, parallel task processing
CollaborateπŸ’¬Free collaboration, shared contextBrainstorming, complex problem solving

Selection Decision Tree​

Start Model Selection
    β”‚
    β”œβ”€β”€ Does task have fixed order?
    β”‚   └── Yes β†’ Pipeline (Sequential Model)
    β”‚
    β”œβ”€β”€ Need content-based task routing?
    β”‚   └── Yes β†’ Route (Routing Model)
    β”‚
    β”œβ”€β”€ Can tasks be parallelized?
    β”‚   └── Yes β†’ Coordinate (Coordination Model)
    β”‚
    └── Need free discussion?
        └── Yes β†’ Collaborate (Collaboration Model)

πŸ”„ 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 β†’ Bot 1 β†’ Bot 2 β†’ Bot 3 β†’ Bot 4 β†’ Final Result

Core Features​

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

Use Cases​

  • Software Development Workflow: Requirements Analysis β†’ Design β†’ Implementation β†’ Code Review β†’ Testing
  • Content Production Pipeline: Content Creation β†’ Editing β†’ Formatting β†’ Quality Review
  • Data Processing Pipeline: Data Collection β†’ Data Cleaning β†’ Data Transformation β†’ 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 β†’ Leader Bot (Routing Decision)
              β”œβ”€β”€ Frontend Issue β†’ Frontend Bot
              β”œβ”€β”€ Backend Issue β†’ Backend Bot
              β”œβ”€β”€ Database Issue β†’ Database Bot
              └── DevOps Issue β†’ DevOps Bot
                      ↓
                   Result

Core Features​

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

Use Cases​

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

πŸ‘₯ 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 β†’ Leader Bot (Task Decomposition)
              β”œβ”€β”€ Analyst Bot ──┐
              β”œβ”€β”€ Data Bot ─────┼→ Leader Bot (Result Aggregation) β†’ Comprehensive Result
              └── Research Bot β”€β”˜

Core Features​

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

Use Cases​

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

πŸ’¬ 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 β†’ Shared Context Space
              ↕        ↕        ↕        ↕
           Bot 1    Bot 2    Bot 3    Bot 4
              β””β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”˜
                         ↓
                Collaborative Output

Core Features​

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

Use Cases​

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

Model Comparison​

Feature Comparison Table​

FeaturePipelineRouteCoordinateCollaborate
Execution OrderFixed sequenceDynamic single pathParallel executionNo fixed order
Bot InteractionUnidirectional passingLeader β†’ ExpertLeader coordinationFull interaction
Context SharingPartial (sequential)Independent contextLeader aggregationFully shared
Team Size3-6 Bots1 Leader + N experts1 Leader + 3-5 experts3-8 Bots
Execution EfficiencyMedium (serial)High (single path)High (parallel)Medium (discussion)
Ease of Use⭐⭐⭐⭐⭐⭐⭐⭐⭐⭐⭐⭐⭐⭐⭐
Flexibility⭐⭐⭐⭐⭐⭐⭐⭐⭐⭐⭐⭐⭐⭐⭐

Selection Guide​

  • Standardized Process β†’ Pipeline
  • Need Specialized Division (Single Expert) β†’ Route
  • Need Specialized Division (Multiple Experts) β†’ Coordinate
  • Need Free Discussion β†’ 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​

ModelKey Points
PipelineControl step count (≀6), single responsibility per step
RouteClear routing rules, include default handling path
CoordinateLeader needs strong aggregation capability, control parallel count (3-5)
CollaborateDefine clear goals, set discussion termination conditions

Choosing the right collaboration model is key to success! πŸš€