ArchiveNET

World's First Decentralized Context Protocol for AI Agents

Comment 1

ArchiveNET

Inspiration

Modern AI agents like ChatGPT, Claude, Cursor, VS Code extensions, and taskbots operate in complete isolation. Each one stores your conversations and context separately, often in corporate-controlled databases, meaning:

  • Every conversation starts from zero context
  • Your memory is fragmented across platforms
  • You don’t own or control your data
  • Important AI knowledge disappears when a platform changes or shuts down

This creates massive friction for users who rely on multiple AI tools in their workflows — whether for development, productivity, research, or creativity. We wanted to eliminate that fragmentation and give users ownership of their AI context and memory.

What it does

  • Maintains a universal, persistent context across all your AI agents
  • Stores context data on-chain, ensuring durability, transparency, and decentralization
  • The context is encrypted and access-controlled, meaning only the user can access or manage their memory
  • Replaces traditional vector databases with an on-chain alternative, removing centralized dependencies
  • Provides APIs for AI agents to fetch, update, or sync contextual memory with user permission

How we built it

  • Developed a lightweight context schema that can be serialized and stored on-chain efficiently
  • Built smart contracts on Arweave to store encrypted AI context
  • Created a Node.js + Express API that AI tools can integrate with to read/write user context
  • Built a basic MCP Server that can be integrated with Claude and Cursor
  • Added support for wallet authentication (e.g., MetaMask) to verify ownership and authorize access

Challenges we ran into

  • Gas efficiency: Storing even small amounts of context data on-chain was initially expensive
  • Encrypted data access: Balancing privacy with usability required careful client-side decryption
  • Tool integration: Many AI tools don’t expose easy APIs for memory access or external hooks
  • Sync conflict resolution: Ensuring consistency when multiple agents access/update memory simultaneously
  • UX tradeoffs: Keeping the on-chain flow seamless for users while ensuring full control

Accomplishments that we're proud of

  • Built a functional prototype of decentralized AI memory management
  • Successfully stored and retrieved encrypted AI memory on-chain
  • Integrated the memory layer with two popular tools (Claude and Cursor MCP Server)
  • Designed a developer-friendly SDK for AI agents to integrate with ArchiveNET
  • Created a user dashboard for reviewing and managing stored context

What we learned

  • Decentralized memory management is possible — but requires thoughtful tradeoffs in storage, latency, and encryption
  • AI tools benefit immensely from shared, persistent context, especially in multi-agent workflows
  • Web3 and AI can intersect meaningfully when privacy, control, and transparency are priorities
  • Building for users who actually want to own their data opens new design paradigms
  • Real-world AI needs universal memory, not siloed silos — and we can build it

What's next for ArchiveNET

  • Integrate with more AI platforms (eg. Gemini, open-source agents)
  • Optimize on-chain storage using rollups or L2 solutions for scalability
  • Implement AI-native memory compression and TTL logic
  • Launch community-owned governance for memory schema evolution
  • Release ArchiveNET SDK v1.0 for public use
  • Explore integration with Web3 identity systems like ENS or Lens
  • Build an open marketplace for AI agents to register and interact with user-controlled context

Built With

Share this project:

Updates