NetworkMind
AI powered relationship intelligence and contact research prototype
NetworkMind combines professional contact data, public web research and LLM assisted responses to help users search their network and understand which people may be relevant to a question.
Overview
NetworkMind is a two person side project and full stack prototype. It combines contact data, public web research and LLM assisted responses to help users understand which people in a professional network are relevant to a given question.
Why We Built It
Network research often starts with scattered contact lists, incomplete notes and manual web searches. The prototype explores whether a guided AI workflow can make that search more focused while keeping context visible.
My Contribution
- Prompt routing and tool selection.
- Retrieval and public research logic.
- Contact ranking and fallback handling.
- Validation, testing and full stack implementation.
How It Works
- Import professional contact data.
- Interpret the user's query.
- Decide whether to search contacts, run public web research or combine both.
- Rank relevant people and return an answer with context.
Technical Architecture
The prototype uses Next.js and TypeScript for the application, Prisma with PostgreSQL for structured data, Redis for supporting application state and REST integrations for AI and search APIs.
Current State
NetworkMind is a prototype, not a finished commercial product. The current work focuses on the search, routing, ranking and response paths that make the idea testable.
What I Learned
I learned how important it is to design AI workflows as systems. Routing, failure handling, testing and clear context around the answer all shape whether the workflow feels useful.