Memo is not just another AI chatbot wrapper. It is an orchestrated multi-agent system where three specialized AI agents — a Researcher, a Summarizer, and a Critic — collaborate through an iterative pi
Memo is an autonomous, multi-agent AI research pipeline designed to run on your own hardware while delivering high-quality, "peer-reviewed" answers to complex questions.
It isn't just a chatbot; it's a structured reasoning engine that uses a cooperative team of AI agents to think, debate, and remember.
When you ask Memo a question, it triggers a sophisticated, multi-stage workflow:
The Researcher: Uses native tool-calling to browse the web (via DuckDuckGo) and search its own "Memory Vault" to gather raw facts.
The Summarizer: Synthesizes the researcher's findings into a structured, concise answer with key takeaways.
The Critic: Acts as a quality control layer. It looks for gaps, logic holes, or missing information. If the answer isn't perfect, it rejects the response and sends the Researcher back out for more data (Iterative Refinement).
Privacy-First (Local LLMs): All the "thinking" happens on your local PC via Ollama. Your research data, private keys, and thoughts never leave your machine.
Semantic Memory: Memo doesn't just store files; it uses Vector Embeddings (via Llama 3) to "remember" the context of past sessions. If you ask a related question later, it recalls what it already knows.
Hybrid Cloud/Local Deployment: Through the Cloudflare Tunnel and Redis refactor we just completed, Memo can live on the web (Vercel) while still using your local GPU for the heavy lifting.
Decentralized Storage (0G Network): Memo is integrated with the 0G blockchain, allowing you to store your research memories in a tamper-proof, content-addressed, decentralized layer.
The frontend is a modern, real-time environment featuring:
Live Agent Visuals: Cartoon characters that physically "walk" and interact on a chalkboard as the agents work.
NDJSON Streaming: You see the agents' "thoughts" and stage updates in real-time as they happen.
Interactive Console: A professional, log-style output that shows the raw data behind the agent's decisions.
In short: Memo turns your local AI into a professional research team that gets smarter every time you use it.
Till now In my college the students are very excited to see open source models doing their reserch work so gracefully and also the memory storage part gives them an edge. A senior of my college had a conversation but then later when he went up to pick the project from the point he felt he jsut loaded it into memory and boom! he could just pic up from where he left
No fundraising whatsoever