A B2B trust marketplace: clients find reliable contractors, and vendors get qualified leads.
Refer
About the project
Refer is a web platform designed to connect clients and IT vendors quickly and reliably. Clients describe their task: whether it’s building a website, developing a mobile app, or implementing a CRM system, - and Refer’s algorithm matches them with the most relevant vendors.
The platform’s uniqueness lies in:
- Algorithmic matchmaking based on onboarding and company profiles.
- MyRefScore - a proprietary reliability and relevance rating for vendors.
- Economy around paid matches - vendors pay to access qualified client requests.
- Seamless integrations with LinkedIn and Zoom.
The initiative came from a large company specializing in lead generation, which wanted to create a more transparent and efficient marketplace for IT services.
Web platform development
Refer
/
/
- Clients: Businesses and individuals in need of IT services - from development to support and audits.
- Vendors: IT companies and agencies looking for new clients.
Target audience
Problems solved:
- For clients - no more long tenders, just fast access to trusted vendors.
- For vendors - a flow of relevant leads and automatic matching to requests.
Value created:
- Clients save time and reduce risks when selecting contractors.
- Vendors increase their conversion rates through targeted leads.
- The system motivates both sides: clients get reliable results, vendors secure new deals.
Key features include:
How it works
- Registration and onboarding for both clients and vendors.
- Creating and publishing service requests.
- Algorithmic recommendations powered by ML.
- Matches & Meetings - direct vendor-client connections, with Zoom API integration.
- Built-in chat for communication.
- Vendor profiles and case studies.
Business model: Vendors pay for access to requests. Pricing is dynamic and depends on the potential deal size.
Payments: Integrated via Stripe.
Loyalty system: MyRefScore ratings and referral mechanics (planned for future releases).
Development process
The project was built entirely from scratch - from architecture to final design.
Team: Backend developers, ML engineers, frontend developers, designer, QA.
Tech stack:
- Frontend: React, TypeScript
- Backend: Python (FastAPI), PostgreSQL
- ML: Matching algorithms and request value estimation
- Infrastructure: Docker, Kubernetes, AWS (CI/CD)
Key challenges solved:
- LinkedIn API restrictions - overcame by combining official integration with internal data enrichment.
- Secure payments - implemented via Stripe Elements, fully PCI DSS compliant.
- Request cost estimation - solved through an ML model trained on historical data.
Refer has already attracted its first paying matches and continues to grow its base of clients and vendors.
Results
Achievements so far:
- First successful transactions between clients and vendors.
- Active testing of chats and meeting functionality.
- Positive pilot feedback praising the ease of vendor discovery and the transparency of the system.
Impact: Refer is positioning itself as a “B2B trust marketplace for IT services.” Even at MVP stage, the platform is proving demand for algorithmic matchmaking in IT outsourcing.
Refer has validated the concept: businesses want a faster, safer way to find IT partners, while vendors benefit from qualified leads and transparent economics.
Would you like the same? Let's connect and explore the world of digital possibilities together.
