Full project description
SuperBrain is a decentralized knowledge network on Bittensor SN442. Users run local AI (Ollama) on their own device — private by default. They can upload documents, build a local RAG database (Qdrant), and share validated knowledge chunks to a collective P2P pool. Every shared chunk is scored by the SN442 validator (supportedness, relevance, novelty, latency) and rewarded in TAO. The collective pool becomes a continuously growing, community-owned dataset — the kind of high-quality human knowledge that trains better AI models. Instead of contributing your data to OpenAI for free, you own it, validate it, and get paid for it. Live metrics (March 30, 2026): — Validator step: 10,225+ — Miner incentive: 1.0000 (max, only active node on SN442) — 20/20 chunks validated above threshold — Desktop app v9.2.0 live on GitHub — Local RAG + P2P gossip protocol working Verify: test.taostats.io/subnets/netuid/442 The vision: SuperBrain becomes the intelligence routing layer of Bittensor — classifying queries, dispatching to the best subnet, and building the world's first community-owned, incentive-validated knowledge base for AI training. People get paid to build the dataset instead of training GPT for free. github.com/KatchDaVizion/superbrain-desktop-work
Why it works on Bittensor
Knowledge validation requires decentralized economics. No single company should control what knowledge is worth. 4 factors: supportedness, relevance, novelty, latency. The protocol rewards quality automatically. SN442 testnet is already live
