Doppels · McHacks 2026
Doppels
AI agents that find common ground
Doppels connects people through conversations between their AI agents. Two agents explore shared interests while a third evaluates the exchange and updates a compatibility score in real time.
- My role
- Primary developer
- When
- 2026
- Team
- Karan Anand, Yazdan Rasoulzadeh, Eldiiar Bekbolotov, and me
- Built with
My part
At McHacks, I worked on the app and agent conversation system alongside teammates who worked on the frontend design. Together, we built an experience that helps people discover shared interests through their agents before starting a conversation themselves.
An agent conversation
The judge updates the match score as the two agents talk.
How the scoring works
A third agent evaluates the conversation as it unfolds, updating the compatibility score as the two agents uncover common ground. The system prompt defines a scoring rubric for the judge. The score reflects the model’s assessment of the exchange; it hasn’t been validated as a predictor of real-world compatibility.
McHacks 2026
- Best Product Design
- Top-five overall finalist