MultiPairAI
Smart AI agents for faster supplier matches
Created on 7th September 2025
•
MultiPairAI
Smart AI agents for faster supplier matches
The problem MultiPairAI solves
MultiPairAI: AI-Assisted Supplier Sourcing System
This application revolutionizes the traditional supplier sourcing process by automating the initial "first touch" with potential suppliers, dramatically reducing the time and effort required for procurement teams to find and evaluate vendors.
What People Can Use It For
Voice-Driven Procurement Requests
- Natural Language Processing: Users simply speak their purchasing needs into the system (e.g., "I need 50 ergonomic office chairs for our new office space")
- Real-time Transcription: Advanced speech-to-text technology converts voice input into structured procurement requests
- Multi-language Support: Handles various languages and accents for global procurement teams
Automated Supplier Discovery
- Intelligent Search: AI-powered search across multiple supplier databases (Europages, Alibaba, ThomasNet, Global Sources, Kompass)
- Smart Filtering: Automatically categorizes and filters suppliers based on requirements
- Contact Information Extraction: Pulls phone numbers, emails, and locations from supplier profiles
AI-Powered Supplier Outreach
- Automated Phone Calls: The system places calls to discovered suppliers using Twilio integration
- Intelligent Conversation: AI agents conduct natural conversations with suppliers to gather:
- Product availability and specifications
- Pricing information
- Delivery timelines
- Quality certifications
- Minimum order quantities
- Real-time Processing: Live transcription and analysis of supplier responses
Knowledge Graph Intelligence
- Structured Data Storage: All conversations and supplier information are stored in a Neo4j knowledge graph
- Relationship Mapping: Builds connections between suppliers, products, and requirements
- Learning Capability: System learns from past interactions to improve future sourcing
Intelligent Supplier Comparison
- Automated Scoring: AI evaluates suppliers based on multiple criteria (price, quality, delivery, ratings)
- Side-by-side Comparison: Visual dashboard showing supplier strengths and weaknesses
- Recommendation Engine: AI suggests the best supplier based on your specific needs
How It Makes Existing Tasks Easier & Safer
Time Savings
- Eliminates Manual Research: No more hours spent searching through supplier directories
- Reduces Call Time: AI handles initial qualification calls, saving procurement teams 80% of their time
- Instant Results: Get supplier responses within minutes instead of days
Risk Reduction
- Consistent Data Collection: Standardized questions ensure no important information is missed
- Audit Trail: Complete record of all supplier interactions and decisions
- Quality Assurance: AI verification of supplier claims and capabilities
Cost Optimization
- Better Price Discovery: AI agents negotiate and compare prices across multiple suppliers
- Reduced Procurement Costs: Eliminates need for dedicated sourcing specialists for routine purchases
- Volume Discounts: AI can identify opportunities for bulk purchasing
Scalability & Global Reach
- 24/7 Operation: Works around the clock across different time zones
- Multi-language Support: Handles suppliers in different countries and languages
- Unlimited Capacity: Can process hundreds of supplier inquiries simultaneously
Real-World Use Cases
Office Equipment Procurement
- "Find me 20 standing desks under $500 each with 5-year warranty"
- System calls 15+ furniture suppliers, compares offers, presents top 3 options
IT Hardware Sourcing
- "I need 100 laptops for new employees, budget $800 per unit"
- AI contacts computer suppliers, gathers specs, negotiates bulk pricing
Services Procurement
- "Find cleaning services for our 5,000 sq ft office in downtown"
- System identifies local cleaning companies, gets quotes, checks references
Manufacturing Components
- "Source custom metal brackets for our product line"
- AI finds specialized manufacturers, discusses specifications, gets samples
Challenges I ran into
I built this in not 2 weeks but in 3 days playing due to limited time i unable to complete my define task but i do my best
Major Technical Hurdles & How I Overcame Them
Real-time Audio Streaming Architecture
The biggest nightmare was getting WebSocket audio streaming to work seamlessly between frontend and backend. Audio chunks were arriving out of order, causing transcription failures and a terrible user experience. I initially tried sending audio chunks immediately as they arrived, but this created race conditions where the backend was processing chunks in random sequence. After hours of debugging, I implemented proper chunk sequencing with index tracking and added a buffering mechanism to handle network latency. I also created a fallback to batch processing when real-time streaming fails, which saved the day when users had poor internet connections.
Knowledge Graph Integration Nightmare
Integrating Neo4j with the Graphiti framework was way more complex than I anticipated. The connection pooling was failing, queries were timing out, and I kept getting cryptic error messages about constraint violations. The documentation was sparse and I had to dig through GitHub issues to understand how to properly initialize the graph database with indices and constraints. I eventually figured out that I needed to run the build_indices_and_constraints function before any other operations, and that the connection had to be properly closed after each operation to prevent memory leaks.
The 3-Day Sprint Reality
Building this entire system in 3 days meant I had to make some brutal trade-offs. I skipped prope
Tracks Applied (4)
Top 100 Participants
Lovable
All Registered Participants
Windsurf
Vibecoding your Idea | Powered by Rocket.new
Rocket
Voice AI & Conversational Products | Powered by Elevenlabs
ElevenLabs
Technologies used