Partnership Agent

An AI-powered partnership management system featuring advanced citation functionality and high-performance vector search that provides precise document references when answering questions about partnership agreements, revenue sharing, compliance, and operational requirements.

Built with Azure OpenAI, Azure AI Search (or Elasticsearch), ASP.NET Core, and enhanced with intelligent document citation tracking and sub-second document retrieval.

🚀 Quick Start (Recommended)

⭐ Preferred Method: Cross-Platform .NET Setup

Works identically on Windows, macOS, and Linux:

# Linux/macOS - Default (InMemory chat history, Elasticsearch)
cd setup
./setup.sh

# Linux/macOS - High-performance vector search with Azure AI Search
cd setup
./setup.sh inmemory --vector-search

# Linux/macOS - SQLite chat history with vector search
cd setup
./setup.sh sqlite --vector-search

# Linux/macOS - Azure SQL chat history  
cd setup
./setup.sh azuresql

# Windows Command Prompt
cd setup
setup.cmd [inmemory|sqlite|azuresql] [--vector-search]

# Windows PowerShell  
cd setup
.\setup.cmd [inmemory|sqlite|azuresql] [--vector-search]

Why this is preferred:

✅ Single codebase - no platform differences
✅ Better error handling and logging
✅ Consistent behavior across all operating systems
✅ Professional development practices
✅ Easier to maintain and extend
🚀 High-performance vector search option for production workloads

🔧 Alternative Setup Methods

Platform-Specific Scripts (Legacy)

Linux/macOS (Bash):

./setup/start-partnership-agent.sh

Windows (PowerShell):

.\setup\start-partnership-agent.ps1

Manual .NET Execution

cd setup
dotnet run --project setup.csproj [inmemory|sqlite|azuresql] [--vector-search]

# Examples:
dotnet run                           # Uses InMemory (default) + Elasticsearch
dotnet run inmemory --vector-search  # Uses InMemory + Azure AI Search (high performance)
dotnet run sqlite                    # Uses SQLite + Elasticsearch
dotnet run sqlite --vector-search   # Uses SQLite + Azure AI Search (recommended)
dotnet run azuresql                  # Uses Azure SQL Database

🔍 Search Engine Options

The Partnership Agent supports two search engine configurations:

🚀 Azure AI Search (Vector Search) - Recommended for Production

Best for: Production deployments, high-performance requirements
Performance: Sub-second document retrieval (99%+ faster than Elasticsearch)
Technology: Vector embeddings with semantic similarity search
Setup: Requires Azure AI Search service configuration
Cost: ~$250/month for Basic tier
Benefits: Semantic understanding, automatic scaling, managed service

🗃️ Elasticsearch (Traditional Text Search) - Default

Best for: Development, testing, offline scenarios
Performance: 60-120 seconds for document search
Technology: Text-based search with keyword matching
Setup: Automatic Docker container deployment
Cost: Free (local Docker container)
Benefits: No cloud dependencies, familiar technology

💾 Chat History Options

The Partnership Agent supports three chat history storage options:

🧠 InMemory (Default)

Best for: Development, testing, quick demos
Persistence: No persistence - data lost on restart
Setup: Zero configuration required
Performance: Fastest startup and response times

🗄️ SQLite (Recommended for Local Development)

Best for: Local development with persistence, offline work
Persistence: Full persistence in local SQLite database
Setup: Automatic Docker container with volume mounting
Performance: Fast local file-based storage
Container: partnership-agent-sqlite with persistent volume

☁️ Azure SQL Database

Best for: Production deployments, multi-user scenarios
Persistence: Full persistence in cloud database
Setup: Requires Azure SQL Database and connection configuration
Performance: Network-dependent, highly scalable
Features: Enterprise-grade reliability and backup

🎯 What Gets Set Up

The setup process configures a complete partnership management system with citation-enhanced AI responses:

🗃️ Enhanced Document Index

8 comprehensive documents with realistic partnership content
Rich metadata including source paths, modification dates, and versioning
Citation-optimized content with specific percentages, requirements, and procedures

📋 Document Categories:

Partnership Agreement Template - Formation, revenue tiers, contribution requirements
Revenue Sharing Guidelines - Detailed calculation methods, payment terms
Partnership Compliance Requirements - Documentation standards, audit procedures
Standard Partnership Contract - IP rights, liability, termination procedures
Partner Onboarding Process - Assessment criteria, training requirements
Dispute Resolution Framework - Escalation procedures, mediation processes
Performance Metrics and KPIs - Revenue metrics, operational standards
Data Security and Privacy Policy - Classification levels, security controls

🔍 Citation Features

Precise text excerpts with character-level positioning
Relevance scoring based on query-answer term matching
Context awareness with before/after text snippets
Multi-source references from multiple documents
Document metadata including categories and modification dates

🧪 Testing Citation Functionality

Quick Citation Tests

# Linux/macOS
./setup/test-citations.sh

# Windows
.\setup\test-citations.ps1

Example Citation Queries

Try these prompts to see rich citations in action:

"What are the revenue sharing percentages for different partner tiers?"
- Expected citations from Partnership Agreement Template and Revenue Sharing Guidelines
- Should show specific percentages: Tier 1 (30-35%), Tier 2 (20-25%), Tier 3 (10-15%)
"What is the minimum credit score requirement for partner verification?"
- Expected citation from Partnership Compliance Requirements
- Should reference the 650 minimum credit score
"How long is the notice period for partnership termination?"
- Expected citation from Standard Partnership Contract
- Should reference the 90-day written notice requirement

🏗️ Architecture Overview

┌─────────────────┐    ┌──────────────────┐    ┌─────────────────┐
│   Console App   │    │     Web API      │    │   Elasticsearch │
│                 │◄──►│                  │◄──►│                 │
│ - Interactive   │    │ - REST endpoints │    │ - Document      │
│   chat          │    │ - Step           │    │   indexing      │
│ - Test prompts  │    │   orchestration  │    │ - Search        │
└─────────────────┘    └──────────────────┘    └─────────────────┘
                                │
                                ▼
                       ┌──────────────────┐
                       │   Citation       │
                       │   Engine         │
                       │                  │
                       │ - Text analysis  │
                       │ - Excerpt        │
                       │   extraction     │
                       │ - Relevance      │
                       │   scoring        │
                       └──────────────────┘
                                │
                                ▼
                    ┌──────────────────────────┐
                    │    Chat History          │
                    │    Storage               │
                    ├──────────────────────────┤
                    │ 🧠 InMemory (Default)    │
                    │ 🗄️ SQLite Container      │
                    │ ☁️ Azure SQL Database    │
                    └──────────────────────────┘

🔌 Running the Applications

After setup, you can run the applications individually:

Web API

cd src/PartnershipAgent.WebApi
dotnet run --urls="http://localhost:5001"

REST API available at http://localhost:5001
Swagger UI at http://localhost:5001/swagger
Health check at http://localhost:5001/api/chat/health

Console Application

cd src/PartnershipAgent.ConsoleApp
dotnet run

Interactive chat interface
Type questions about partnerships
See detailed citations in responses
Type 'quit' to exit

Example API Request

curl -X POST "http://localhost:5001/api/chat" \
  -H "Content-Type: application/json" \
  -d '{
    "threadId": "test-123",
    "prompt": "What are the revenue sharing percentages?"
  }'

📊 Citation Response Format

Responses now include detailed citations:

{
  "answer": "Revenue sharing is structured in tiers...",
  "confidence_level": "high",
  "source_documents": ["Revenue Sharing Guidelines", "Partnership Agreement Template"],
  "citations": [
    {
      "document_id": "doc2",
      "document_title": "Revenue Sharing Guidelines",
      "category": "guidelines",
      "excerpt": "Tier 1 (Strategic Partners): 30% of net revenue, paid monthly",
      "start_position": 1450,
      "end_position": 1502,
      "relevance_score": 0.92,
      "context_before": "Partner shares are distributed according to contribution tiers:",
      "context_after": "- Tier 2 (Operational Partners): 20% of net revenue"
    }
  ],
  "has_complete_answer": true,
  "follow_up_suggestions": [...]
}

🛠️ Requirements

Essential Requirements

.NET 8 SDK - Download here
Docker - Docker Desktop (Windows) or Docker Engine (Linux/macOS)
- Required for Elasticsearch (default search)
- Required for SQLite chat history (optional)
Azure OpenAI - Access to Azure OpenAI service with deployed model
Internet connection - For downloading Docker images

Optional Requirements (Based on Configuration)

Azure AI Search - Required for high-performance vector search (--vector-search option)
Azure SQL Database - Only required if using Azure SQL chat history option

For High-Performance Vector Search

When using --vector-search option, you'll need:

Azure AI Search service (Basic tier recommended)
Azure OpenAI with embeddings model (text-embedding-ada-002)
Additional Azure costs (~$250/month for Basic tier)

📚 Documentation

Detailed guides available in the /docs directory:

Core Setup & Configuration

Setup Instructions - Complete manual setup guide
Azure OpenAI Setup - Azure OpenAI configuration
Azure AI Search Setup - 🚀 High-performance vector search configuration
Azure Key Vault Setup - Enterprise credential management
Credential Management - Security best practices
Chat History Configuration - Chat persistence options

System Features & Architecture

Evaluation System - 📊 AI response evaluation framework with quality metrics
Observability - Monitoring and telemetry configuration
Performance Optimization - Performance tuning strategies

Testing & Troubleshooting

Citation Testing Guide - Comprehensive citation testing
Timeout Troubleshooting - Network and timeout issues
No Timeouts Configuration - Disable timeout configurations

🧩 Project Structure

partnership-agent/
├── src/
│   ├── PartnershipAgent.Core/          # Core business logic & citation engine
│   ├── PartnershipAgent.WebApi/        # REST API endpoints
│   ├── PartnershipAgent.ConsoleApp/    # Interactive console interface
│   └── PartnershipAgent.Core.Tests/    # Unit tests including citation tests
├── setup/                              # Setup scripts and tools
│   ├── setup.sh / setup.cmd           # 🌟 Cross-platform setup (PREFERRED)
│   ├── start-partnership-agent.sh     # Legacy bash script
│   ├── start-partnership-agent.ps1    # Legacy PowerShell script
│   ├── test-citations.sh/ps1          # Citation testing scripts
│   ├── CrossPlatformSetup.cs          # .NET setup application
│   └── sample-documents-bulk.json     # Enhanced sample documents
└── docs/                               # Documentation

🔍 Key Features

Citation Engine

Smart text analysis with relevance scoring
Precise positioning with character-level accuracy
Context extraction for better understanding
Multi-document synthesis from various sources

Document Processing

Enhanced metadata including source paths and timestamps
Categorized content (templates, guidelines, policies, contracts)
Structured indexing optimized for citation extraction

AI Integration

Azure OpenAI for natural language understanding
Semantic Kernel for agent orchestration
Step-based processing with entity resolution and response generation

⚙️ Configuration Options

Chat History Provider Configuration

You can configure chat history storage through user secrets or directly:

# Set chat history provider
dotnet user-secrets set "ChatHistory:Provider" "sqlite"

# Configure SQLite connection (if using SQLite)
dotnet user-secrets set "SQLite:ConnectionString" "Data Source=/data/partnership-agent.db;Cache=Shared"

# Configure Azure SQL connection (if using Azure SQL)
dotnet user-secrets set "AzureSQL:ConnectionString" "your-azure-sql-connection-string"

Runtime Chat History Switching

You can also switch providers at runtime by modifying appsettings.json:

{
  "ChatHistory": {
    "Provider": "sqlite"  // "inmemory", "sqlite", or "azuresql"
  },
  "SQLite": {
    "ConnectionString": "Data Source=/data/partnership-agent.db;Cache=Shared"
  }
}

🚦 Getting Started Steps

Clone the repository
Choose your configuration:
- ./setup/setup.sh (InMemory + Elasticsearch - fastest setup)
- ./setup/setup.sh sqlite (SQLite + Elasticsearch - persistent, local)
- ./setup/setup.sh sqlite --vector-search (SQLite + Azure AI Search - high performance)
- ./setup/setup.sh azuresql (Azure SQL - production-ready)
Configure Azure AI Search (if using --vector-search):
- See Azure AI Search Setup Guide
Test with sample queries about revenue sharing, compliance, or termination procedures
Explore the citation responses to see precise document references
Try the console app for interactive testing with persistent chat history

🤝 Contributing

When contributing to this project:

Use the cross-platform .NET setup for consistency
Run the citation tests to ensure functionality
Follow the existing code patterns for citation extraction
Update documentation for any new citation features

📄 License

This project is provided as sample code for educational and demonstration purposes.

Name		Name	Last commit message	Last commit date
Latest commit History 31 Commits
data		data
docker		docker
docs		docs
setup		setup
src		src
.gitignore		.gitignore
LICENSE		LICENSE
MIGRATION_PLAN.md		MIGRATION_PLAN.md
PartnershipAgent.sln		PartnershipAgent.sln
README.md		README.md
console-test.sh		console-test.sh
test-ground-truth.cs		test-ground-truth.cs

License

bobjac/partnership-agent

Folders and files

Latest commit

History

Repository files navigation