🧬 Parallel Me
Inspiration
We often imagine “what if” versions of ourselves — the one who studied differently, built a project sooner, or took that risk.
But what if you could watch those versions learn in real time?
It’s not just an app — it’s a living reflection of your curiosity, evolving in parallel.
Parallel Me was born from that idea: to turn reflection into simulation.
Instead of passively journaling or reading, users create AI twins that study, explore, and evolve alongside them — learning different things, in different ways.
It’s the next step in human learning: not just thinking, but seeing yourself think differently.
What it does
Parallel Me builds a network of parallel learning agents — intelligent twins that represent alternate versions of you.
Each one:
- Learns new topics or skills you assign (or it discovers on its own)
- Reflects on its learning style and progress
- Shares summaries, insights, or creative outcomes back with you
- Evolves its perspective based on your feedback and real-world habits
The result is a parallel learning ecosystem: you can watch how different versions of “you” explore ideas — like one twin diving into psychology papers while another learns coding or design.
They’re not just mirrors — they’re co-learners, discovering new things on your behalf.
How we built it
- 🧠 Multi-agent architecture — each twin is a persistent autonomous agent with its own memory, goals, and reasoning style with Letta AI, and Elevenlabs.
- ⚙️ Supabase for real-time state updates, vector memories, and per-user isolation.
- 💬 Next.js + React Native frontend, integrated with OpenAI/Anthropic APIs for agent cognition and reflection.
- 🔁 Active-learning loops — twins generate knowledge, validate it, and evolve their behavior based on user reinforcement.
- 🧩 Built deploy-first via Vercel, scalable to both mobile and web.
Challenges we ran into
- Designing agent individuality: how to make each twin think and learn differently while staying “you.”
- Handling knowledge overlap between twins to avoid redundancy.
- Creating trustworthy learning feedback — so users feel empowered, not replaced.
- Visualizing parallel learning timelines in a way that’s intuitive and rewarding.
Accomplishments that we're proud of
- Built the whole stack from 0!
- Ideating and land the idea in very short duration.
- Learn and research on cognitive science papers, learning theories, and newest AI development.
- Demonstrated some early promise for accelerated human learning through parallel cognition.
What we learned
- Learning can be decentralized — distributed across intelligent, evolving versions of self.
- Watching a “parallel you” study can reveal your own blind spots and learning biases.
- Reflection becomes richer when it’s externalized through autonomous agents.
- The future of personal AI may not be automation — it’s collaborative cognition.
What's next for Parallel Me
This is a very exciting project for us and we plan to continue iterate on the idea.
- 🧠 Collective memory layer — allow parallels to exchange discoveries, forming a shared knowledge graph.
- 🎥 “Parallel Stream” visualization — see your parallels talk, interact to other agents in real time, like watching a parallel-universe classroom.
- 🤝 Collaborative learning mode — connect your twins with others to co-learn and co-create across domains.
- 🌍 Long-term goal: a new paradigm of learning — the Parallel Study Network, where curiosity scales beyond one brain.
✨ About the project
Your curiosity, multiplied. Watch your digital parallels explore the worlds you imagine.
More than chatbots. Beyond avatars. They’re also learning agents that mirror your curiosity.
Welcome to the age of parallel learning. 🌌
Built With
- fishaudio
- letta
- supabase
- vercel

Log in or sign up for Devpost to join the conversation.