This tutorial demonstrates how to integrate OVADARE with AutoGen to orchestrate multi-agent workflows for preparing meetings while detecting and resolving conflicts.

​
Overview

Dave is preparing for an important sales meeting and sets up multiple agents in AutoGen:
  • Lead Researcher Agent: Conducts research on the target company.
  • Product Specialist Agent: Analyzes product fit for the target.
  • Sales Strategist Agent: Develops a tailored sales strategy.
  • Briefing Coordinator Agent: Compiles a detailed meeting briefing.

​
Workflow

The agents are tasked to:
  1. Research the target company’s background and industry trends.
  2. Analyze product fit based on the research.
  3. Develop a comprehensive sales strategy.
  4. Compile all findings into a detailed briefing document.
Outputs include:
  • Research reports.
  • Industry analysis.
  • Sales strategies.
  • Final briefing documents.

​
Detecting Conflicts with OVADARE

In this scenario, conflicts might arise if:
  • Agents compete for overlapping resources.
  • Strategies contradict each other.
  • Policies regarding sensitive data are violated.

​
Step 1: Setting Up OVADARE

First, ensure OVADARE is integrated with AutoGen: 
from ovadare.conflicts.conflict_detector import ConflictDetector
from ovadare.policies.policy_manager import PolicyManager
from ovadare.resolutions.resolution_engine import ResolutionEngine

conflict_detector = ConflictDetector()
policy_manager = PolicyManager()
resolution_engine = ResolutionEngine()

​
Step 2: Define Policies

Define policies to govern agent actions: 
policy_manager.add_policy({
    'name': 'DataSensitivityPolicy',
    'rules': {
        'access_level': 'restricted',
        'resource': 'confidential_documents'
    }
})

​
Step 3: Monitor Agent Actions

As the agents in AutoGen execute tasks, OVADARE listens for activities: 
agent_actions = [
    {'agent_id': 'lead_researcher', 'action': 'access_data', 'resource': 'confidential_documents'},
    {'agent_id': 'sales_strategist', 'action': 'create_strategy', 'strategy': 'aggressive_pricing'}
]
conflicts = conflict_detector.detect(agent_actions)

​
Step 4: Resolving Conflicts

If conflicts are detected, OVADARE’s Resolution Engine generates recommendations: 
if conflicts:
    resolutions = resolution_engine.generate_resolutions(conflicts)
    for resolution in resolutions:
        print(f"Resolution: {resolution}")

​
Step 5: Applying Resolutions in AutoGen

Send the resolutions back to AutoGen agents: 
for resolution in resolutions:
    agent_id = resolution['agent_id']
    action = resolution['corrective_action']
    autogen_api.apply_resolution(agent_id, action)

​
Outcome

By using OVADARE, Dave ensures:
  • Agents collaborate without policy violations.
  • Conflicts are resolved before disrupting workflows.
  • The meeting preparation process runs smoothly, resulting in a well-prepared team.

​
Next Steps