Over the past ten days, I’ve been building something that started as a small experiment but ended up becoming one of my favorite side projects: Wisdom Chatroom.
I’ve always believed that when facing complex problems, a single answer isn’t enough—what we really need is the collision of ideas. So I asked myself:
What if I could bring several iconic mentors into one virtual room—like Warren Buffett talking about investments, Tim Cook discussing products, and Mark Zuckerberg focusing on innovation—and let them all respond to you in their own voices?
That’s how this project was born.
Start Chatting with AI Mentors
The Vision and Core Features
This chatroom is not just another Q&A bot. It’s a virtual roundtable of mentors:
You ask a question, and it feels like entering a meeting room where several mentors are present.
Each mentor responds in their unique tone, values, and style.
They even “react” to each other’s answers, adding comments on the last two messages, creating a real conversation dynamic.
I want this project to be more than just an LLM application. It’s an experiment in collective intelligence—giving language models personalities and letting them interact like real people, helping users gain insights from multiple perspectives.
Development Journey
This was a true full-stack experience, and I built every layer myself:
Backend: Flask powers the APIs and streaming response endpoints.
LLM Layer: LangChain orchestrates conversations with both OpenAI and Qwen models, supporting multi-mentor responses and reaction prompts.
Frontend: React + Vite builds a clean, chat-like UI, inspired by messaging apps.
Deployment: Nginx serves as the reverse proxy, AWS EC2 hosts the services, and Route53 handles domain + HTTPS.
System & Infra: Linux with firewall tuning and PM2 process monitoring to ensure stability.
I won’t bore you with the technical details here—they’re just tools that turn ideas into reality. What’s truly exciting is watching these “mentors” interact in a way that feels alive.
Why I Built It
I wanted to create something that doesn’t just provide answers, but actually helps users think.
A single LLM is like a super-smart assistant. But when multiple mentors with different “personalities” gather around to discuss your question, it feels more human, more realistic—like a mastermind group sharing their insights.
This is just a prototype, but I believe it has room to grow.
I want to add more personalities—imagine inviting Elon Musk or Steve Jobs into the chatroom—and integrate knowledge retrieval (RAG) so mentors can provide expert, data-driven answers.
Closing Thoughts
This is my favorite project that combines AI + full-stack development from start to finish.
In these ten days, I’ve crafted prompts, fought with CSS bugs, debugged Linux firewalls, and wrestled with AWS Route53 and Nginx configs. But when I finally saw the chatroom come alive—mentors “speaking” on screen—that satisfaction was worth every line of code.
Life Mentor Chatroom is not just a product; it’s an exploration:
What happens when multiple LLMs form a kind of collective intelligence? Could this bring us closer to a truly “thinking” digital companion?
LLM · Flask · LangChain · React · Nginx · AWS · Linux