Inspiration

  • The loan market relies heavily on secure, tamper-proof documentation during origination, trading, and management. Traditional Web2 storage solutions expose critical loan agreements to risks of alteration, loss, censorship, and high recurring costs—creating friction, disputes, and inefficiencies in a multi-trillion-dollar industry. We were inspired by the LMA's mission to drive transparency, efficiency, and sustainability, envisioning a future where loan documents are indestructible, instantly verifiable, and permanently owned without centralized vulnerabilities.

What it does

  • Mneme is a desktop-based prototype that enables lenders and borrowers to upload, AI-steganographically sign, and store loan documents directly on blockchain. It embeds invisible digital signatures and ownership proofs within the files themselves using advanced AI-driven steganography—making documents tamper-proof and verifiable. Users can instantly confirm authenticity and ownership via a simple verification page, eliminating disputes while ensuring permanent, decentralized storage that survives any single-point failure.

How we built it

  • We developed a desktop application prototype allowing seamless file upload and processing. AI-powered steganography invisibly embeds cryptographic signatures and metadata into documents (PDFs, images, etc.). These enhanced files are then stored immutably on blockchain with one-time payment permanence. A clean verification interface extracts hidden proofs and validates ownership on-chain—all designed for commercial viability in loan origination and document management.

Challenges we ran into

  • Storing files directly on blockchain is inherently expensive and slow compared to Web2 solutions—traditional on-chain storage can cost hundreds of dollars per MB due to replication and consensus overhead. Achieving Web2-level cost-effectiveness required innovative permanent storage layers that charge only once while maintaining global redundancy. Integrating invisible AI steganography added complexity: ensuring signatures remain undetectable yet reliably extractable, preserving file integrity, and scaling AI processing for large loan document volumes without compromising speed or security—all while keeping the solution desktop-friendly and commercially scalable.

Accomplishments that we're proud of

  • Delivered true indestructibility: Loan documents become permanently stored and impossible to alter or delete.
  • Invisible yet powerful AI signing: Steganographic signatures are completely hidden to the human eye but instantly verifiable on-chain.
  • Massive efficiency gains: Eliminated Web2 recurring fees and centralized risks while matching cost-effectiveness through one-time permanent storage.
  • Instant verification portal: Reduced document authentication time from days to seconds—critical for loan trading and compliance.
  • Greener footprint: Removed paper-based processes and reduced reliance on energy-intensive data centers.

What we learned

  • We deepened our understanding of the loan market's pain points—disputes over document authenticity, high verification costs, and fragility of centralized storage. Balancing blockchain permanence with Web2 affordability is possible through specialized permanent storage networks. AI steganography proves far superior to traditional digital signatures for ownership proof, offering both security and invisibility. Finally, desktop-based solutions remain essential for enterprise adoption in regulated industries like lending.

What's next for Mneme

  • Enterprise integrations with existing loan origination systems (LOS) and document management platforms.
  • Expand AI steganography to support all loan-related file types (audio recordings, video KYC, scanned agreements).
  • Add multi-party verification workflows for syndicated loans and secondary trading.
  • Compliance modules for regulatory standards (KYC/AML traceability, audit trails).
  • Partnerships with lenders to pilot real-world loan document workflows.

Built With

Share this project:

Updates