Conference-Go
Strategic Sales agents are responsible for making deals in person.
Core trick
Gemini parses attendee descriptions into structured JSON, embeds them, and ranks matches by brute-force similarity search over the vectors.
Description
Strategic Sales agents are responsible for making deals in person. Conference-Go helps Sales teams prepare in advance by finding best conference matches, sending emails, and synthesizing personas to practice pitching before meeting a real person.
README excerpt
# Conference-Go An AI-enabled smart conference connection matching and email generation system using Google Gemini API via AI Studio. ## Features - **Preprocessing Module**: Uses Gemini to parse natural language descriptions of people into structured JSON format optimized for embeddings. - **Embedding Module**: Embeds structured person data into vector space using Gemini embedding models. - **Indexing Module**: Brute-force indexing and similarity search for person embeddings. - **Attendee Profiles**: 308 diverse fake person descriptions for testing and demonstration. - **Connection Matching**: Input an ideal connection description and find the most similar people from the dataset. - **Email Drafting**: AI-generated introduction emails between matched attendees. - **Chat**: Converse with an AI persona of any attendee. - **Text-to-Speech**: Listen to attendee responses via Gemini TTS. ## Quick Start (Docker) ```bash ./setup.sh ``` The script checks for Docker, prompts for your API key (saves to `.env`), builds the image, and starts the server at http://localhost:8080. The SQLite database is mounted as a bind volume so profiles persist across container restarts. Or manually: ```bash export GOOGLE_API_KEY=your_key_here docker compose up --build ``` ## Usage ### Web Interface (Recommended) Run the application in web server mode: ```bash go run cmd/server/main.go -web ``` Or specify a custom port: ```bash go run cmd/server/main.go -web -port 3000 ``` Then open http:/