Close more deals faster by instantly surfacing credible, tailored customer references for your sales conversations.
An intelligent AI-powered tool that helps sales teams find relevant customer quotes, case studies, and reviews. This tool enables Act-On sales representatives to deliver more credible, personalized, and persuasive pitches by seamlessly integrating the voice of the customer into the sales process.
- AI-Powered Matching: Intelligently finds the most relevant customer references based on natural language requests
- Smart Filtering: Filter by industry, market segment, use case, CRM type, and company
- Confidence Scoring: View confidence percentages showing how well each reference matches your request
- Smart Highlighting: Automatically highlights the most relevant parts of each quote
- Intelligent Request Interpretation: Expands abbreviations and optimizes search queries
- Multiple AI Models: Choose between Anthropic Claude, OpenAI, or Google Gemini
- Easy Copying: Copy individual references or all matches with a single click
- Demo Mode: Test functionality instantly without consuming API quota
# Clone the repository
git clone https://github.com/act-on/reference-agent.git
cd reference-agent
# Install dependencies
npm install
# Set up environment variables
cp .env.example .env
# Edit .env to add your API keys
# Start development server
npm run devVisit http://localhost:3000 to view the application.
# Start development server with Turbo
npm run dev
# Build for production
npm run build
# Start production server
npm run start
# Preview production build locally
npm run preview
# Lint code
npm run lint
npm run lint:fix
# Type check
npm run typecheck
# Run both lint and type checks
npm run check
# Format code
npm run format:write
npm run format:check- Enter Your Request: Describe what you're looking for in natural language (e.g., "Find me quotes from healthcare customers about email deliverability")
- Apply Filters (Optional): Click the "Add Filter" button to filter by industry, market segment, use case, or CRM type
- View Results: See matching customer quotes ranked by relevance with key phrases highlighted
- Copy References: Use the "Copy" button on individual quotes or "Copy All" to use in your sales communications
- Toggle Demo Mode: For training or quick testing without consuming API quota
References follow this template:
Quote: [Reference Detail]
Customer Contact: [Customer Contact], Account: [Account Name]
Link to Case Study or G2 Review: [Case Study Link]
Link to Slide: [Reference Slide Link]
- Request Enhancement: The system interprets your request, expanding abbreviations and optimizing search terms
- Filtering: Your query is matched against filtered customer references from approved sources
- AI Ranking: References are scored based on relevance to your request using Claude, GPT, or Gemini
- Result Refinement: A second AI pass identifies key highlights and adjusts confidence scores
- Presentation: Top matches are displayed with highlighted text for easy scanning
This project uses the T3 Stack:
- Next.js - React framework with server components
- TypeScript - Type safety
- Tailwind CSS - Utility-first CSS
- shadcn/ui - UI component system
- Vercel AI SDK - AI capabilities integration
- Anthropic Claude (default): Excellent for nuanced understanding of customer references
- OpenAI GPT-4o: Alternative model with different strengths
- Google Gemini: Additional model option
Create a .env file with the following variables:
# AI API Keys (at least one is required)
ANTHROPIC_API_KEY=your_anthropic_api_key
OPENAI_API_KEY=your_openai_api_key
GOOGLE_GENERATIVE_AI_API_KEY=your_gemini_api_key
FULL_PASS=your_password_for_admin_access
- Analytics integration with Google Sheets
- Enhanced query interpretation to distinguish between quotes, case studies, and reviews
- Salesforce API integration
- Sorting results by match percentage