DocsOnChain
Turning PDFs into Monetizable, Verifiable, AI-Searchable Knowledge with DocsOnChain
Global
Research
6 Team Members
April 2 - July 26
Problem
People have valuable documents (PDF guides, research, internal notes) but turning them into a product is hard: uploading + access control + payments + trustworthy ownership + real “ask my document” AI typically requires stitching together many systems.
DocsOnChain makes documents sellable and shareable, and adds a conversational AI that answers questions using the full context of the user’s uploaded files—so users don’t have to manually search, skim, or copy/paste content.
The Challenge
The core challenge was building an experience that feels simple for users while coordinating multiple guarantees behind the scenes:
Document-grounded answers (RAG) that stay constrained to the uploaded content.
Secure delivery of paid content (one-time download links) and permissioned sharing.
Verifiable provenance + access events, so creators and buyers can trust who owns what and who was granted access.
Low-friction payments tied to a specific resource, including a service fee gate and a per-document price.
Solution
DocsOnChain is a full-stack system with a Next.js frontend and a NestJS backend that orchestrates storage, payments, AI retrieval, and ledger-backed auditability.
When a user uploads a PDF, the backend stores the file in Google Cloud Storage, extracts text, chunks it, generates Gemini embeddings (embedding-001), and indexes the chunks in Pinecone under the document’s ID. This powers fast, high-relevance retrieval for chat.
For chat, the system performs vector search per selected document IDs, builds a constrained prompt that instructs the model to answer only from retrieved chunks, and maintains conversation continuity using Redis-backed history + rolling context.
For monetization and trust, DocsOnChain enforces payments at the API boundary using x402 (USDC on Base network), including a service fee step and a per-file price check. Access grants, favorites, and payment confirmations are written as Hyperledger Fabric transactions (e.g., CreateFile, GrantAccess, ConfirmPayment, ToggleFavorite), creating an auditable record of ownership and entitlements. Users receive content via one-time signed download links to prevent uncontrolled sharing.
Conclusion
DocsOnChain shows how retrieval-based AI, modern payment rails, and ledger-backed audit trails can combine to turn static documents into interactive, monetizable products.
As AI-driven content experiences mature, systems that pair grounded answers with verifiable access + payment enforcement will be the most trusted by creators and users alike.
DocsOnChain
Monetize PDFs with paywalled access, ledger-backed sharing records, and an AI chatbot that answers from your documents’ full context.
