Agora

Home Landing Page
Dashboard
Linkedin Auth + Profile Storage
Find Relevant Connections
Draft Personalized Email / Message
Automatic Email Sent with AgentMail
AgentMail Inboxes for Automated Messages

Inspiration

During this intense recruiting season, we found ourselves drowning in stress while desperately trying to secure internships through every possible avenue, from mass applications to daily coffee chats for referrals. We quickly realized how much excess time we were wasting on the slow and tedious process of finding the right LinkedIn connections, crafting personalized messages for each person, and managing endless back-and-forth scheduling conversations online. We wanted a way to automate that entire process, so job seekers could focus on real conversations instead of logistics. That's why we made Agora, an intelligent automation system that streamlines the entire networking and outreach process.

What it does

Agora is a comprehensive LinkedIn outreach automation platform that transforms cold prospecting into personalized connections at scale. The system intelligently searches LinkedIn for relevant contacts based on user queries, automatically analyzes profile data to rank connections by relevance based on your own LinkedIn profile and information (computes similarity score based on university, student organizations, hometown, etc.), and generates AI-powered personalized messages tailored to each contact’s background and role. AgentMail then automates the whole conversation process, scheduling meetings for you. Users can seamlessly manage their entire outreach workflow (discovery → message drafting → sending) all within a sleek and intuitive dashboard that makes professional networking super effortless and strategically optimized.

How we built it

It’s a full-stack application with a Next.js frontend (Tailwind CSS and Framer Motion for a polished user experience), backed by a Fastify API that orchestrates the entire pipeline. The core data extraction engine leverages Python scripts using StaffSpy and Selenium to scrape LinkedIn profiles in real-time, while our scoring algorithm ranks contacts by relevance to user queries. We integrated generative AI for personalized message creation and built seamless communication between our TypeScript backend and Python processing layer, managed through a Turborepo monorepo structure that keeps our codebase organized.

Frontend (Next.js) → API (Fastify) → Python Scripts (StaffSpy/Selenium) LinkedIn Data → CSV Processing → Scoring Algorithm → Frontend Display Draft Messages → AI/Template Generation → Email/LinkedIn Sending

Challenges we ran into

LOTS of git merge issues
API policies for LinkedIn and other big companies
Debugging
Mac vs Windows dependencies, we each use a different one so difficult to coordinate
Getting distracted by the activities on North Campus

Accomplishments that we're proud of / What we learned

Being the first hackathon for most of us, we learned how to design and ship a complete full-stack AI product super fast, from scraping pipelines to LLM prompt engineering to API integrations. We got hands-on experience with structured AI outputs, backend/frontend orchestration, and handling messy real-world data.