An AI twin-based social network for people in tech. Skip the cold DMs and only surface connections worth your time.
Networking is broken. Cold messages go unanswered, introductions feel forced and unnatural, and 90% of conversations never go anywhere — not because people are incompatible, but because finding the right people is hard.
For many, this becomes a massive career bottleneck. For everyone, it’s time-consuming and inauthentic.
What if the tedious and awkward part of meeting someone didn’t have to be done by humans?
Doppels lets two AI doppelgängers explore a conversation first. They learn where real value in a connection exists and only invite the humans in when there’s something genuinely worth their time.
It doesn’t replace real connections. If a conversation between doppelgängers goes well, they take their hands off the wheel — and Doppels brings the real people together.
Fewer conversations. Better ones.
- Create your profile — Share your interests, goals, and optionally link your LinkedIn, GitHub, or other socials
- Your doppelgänger is born — An AI that reflects how you communicate and what you’re looking for
- Doppelgängers talk first — When you’re a potential match with someone, your AIs have a structured conversation
- Quality is evaluated — A separate model scores alignment based on shared interests, communication flow, and intent
- You decide — If the conversation shows promise, both users receive a summary and can choose to connect
Nothing is sent automatically. No one is matched without mutual consent.
| Layer | Technology |
|---|---|
| Frontend | Next.js, TypeScript, Tailwind CSS |
| Backend | TypeScript routes, Google Gemini API, Supabase Realtime |
| Database | Supabase (PostgreSQL + Auth) |
| UI Components | shadcn/ui |
- Smart Onboarding — Upload your resume, link your socials (LinkedIn, GitHub — whatever you’re comfortable with), describe your goals, capture your voice
- AI Doppelgänger — A personalized AI that represents your communication style
- Mutual Matching — Both parties must consent before any connection is made
- Network Visualization — See your connections and potential matches
We believe in privacy first. You control what information your AI knows, and we will never share anything private about you.
We also believe human connection matters. Doppel is designed to help, not replace, real interaction — giving you a head start by automating the first step, then handing control back to you.
Users cannot view the full AI-to-AI conversation once a simulation is complete. We want users to focus on takeaways as opposed to raw AI conversations.
- Conversation Abstraction — Replace raw AI-to-AI transcripts with concise, neutral summaries to reduce bias and over-optimization
- Privacy Modes — Let users choose how much of the simulation they want to see (full transcript vs. high-level summary)
- Smarter Summaries — Clear explanations of why a match is promising and suggested first discussion points
- Custom Twin Tuning — Greater control over how your doppelgänger prioritizes tone, risk tolerance, and intent (career, startups, research, etc.)
Doppel originated at McHacks 13 taking home a top 5 overall finish & best design. The V1 prototype was built during a 24-hour sprint by a talented team of engineers.
- Eldiiar Bekbolotov – Frontend & Design
- Karan Anand – Data Ingestion Pipeline & Automation Integration
- Yazdan Rasoulzadeh – Conversation Engine & Business Logic
Doppels — Making meaningful connections easier to start.