NeriaAI

Problem Statement

Neria is a decentralized AI reasoning platform that turns AI outputs into permanent, verifiable knowledge assets on-chain through autonomous multi-agent validation. Built for researchers, developers, and professionals who need trustworthy AI reasoning, it uses five collaborating agents to research, reason, and validate answers, then mints each verified result as a Knowledge Capsule NFT on Base L2 with IPFS storage and vector search for discovery. Think: Perplexity meets Stack Overflow, where blockchain preserves and attributes knowledge.The Problem: Millions of researchers, developers, and professionals rely on AI for critical decisions, but face fundamental trust issues:AI provides fluent but unverified outputsStack Overflow and forums have conflicting or outdated answersProfessionals waste critical time manually cross-checking AI results.Most AI answers are disposable, leading to repetitive effort and high information decay.The Solution: 5 specialized uAgents collaborate to research, reason, and validate AI responses through autonomous consensus (2/3 approval required). Each verified answer becomes a Knowledge Capsule NFT, secured on Base L2.How It Works:User asks question via chat interfaceQuery Router Agent classifies and routes the requestResearch Agent gathers context from existing knowledge capsules and webReasoning Agent generates structured logic using MeTTa knowledge representationValidation Agent coordinates 3 sub-validators (Logic, Source, Completeness)Capsule Agent mints Knowledge Capsule NFT on Base L2 with IPFS storageCreator receives attribution for future citationsKey Innovation: Autonomous multi-agents validate reasoning with zero human bottleneck.Impact: Transforms disposable AI answers into permanent, verifiable, and reusable knowledge assets that creators can own and benefit from.

Solution

Multi-Agent Architecture: We built 5 specialized uAgents using Fetch.ai framework - Query Router, Research, Reasoning, Validation, and Capsule agents. Each agent has specific responsibilities and communicates via Chat Protocol. Agent Orchestration: Used Agentverse registry for agent discovery and communication. Reasoning Engine: Integrated SingularityNET's MeTTa for formal logic representation. ASI:One API powers the reasoning generation with structured knowledge graphs. Validation System: Implemented 3-sub-validator consensus (Logic, Source, Completeness) with 2/3 approval requirement. Each validator checks different aspects of reasoning quality. Blockchain Integration: Deployed ERC-721 Knowledge Capsule NFTs on Base L2. Each verified answer becomes a permanent asset with IPFS hash storage. Storage Layer: IPFS (Pinata) for full data storage, Supabase PostgreSQL with pgvector for semantic search. Vector embeddings enable discovery of related knowledge capsules. Frontend: Next.js 15 with React 19, TypeScript, Tailwind CSS, Shadcn UI. Web3 integration via Privy, Viem, Wagmi for wallet connectivity. Achievement: We created autonomous agent consensus without human intervention - agents validate each other's reasoning through structured validation protocols, eliminating the need for human reviewers while maintaining quality.