Store AI-generated facts with cryptographic proof. Verify truth, not just data.
AI models hallucinate. But even when they don't — how do you prove it?
When an AI agent generates a fact, there's no way to verify:
- Was this fact tampered with after it was generated?
- Did the AI actually produce this output?
- Was the fact stored at a specific point in time?
RehmanFactAI stores every AI-generated fact on Shelby — a decentralized verifiable hot storage layer — with a SHA-256 hash and timestamp baked in at write time.
- Cryptographic integrity — SHA-256 hash stored alongside every fact
- Sub-second reads — Shelby hot storage means no cold-start latency
- 70% cheaper egress — vs AWS S3 / Google Cloud Storage
- Tamper detection — Any modification to stored data is immediately detectable
- AI-agent ready — Designed for autonomous agents submitting and verifying claims
- TypeScript-first — Fully typed interfaces
- TypeScript 5.x
- @shelby-xyz/sdk — Official Shelby decentralized hot storage SDK
- Node.js 18+
- crypto (Node built-in) — SHA-256 hashing
git clone https://github.com/huzaifakhan98/RehmanFactAI
cd RehmanFactAI
npm installRehmanFactAI/
├── src/
│ └── index.ts # Core FactStorage class
├── package.json
├── tsconfig.json
└── README.md
Powered by Shelby decentralized hot storage.