ProofSettle

Inspiration

Every damaged shipment can trigger a frustrating dispute between a buyer and a supplier. Buyers need compensation for defective goods, while suppliers need proof that deductions are justified. Evidence is often scattered across photographs, purchase orders, invoices, and emails, making it difficult to reach a fair resolution. We wanted to build a system that connects what both parties agreed to before shipping with the evidence collected when goods arrive. What it does

ProofSettle is a contract-aware settlement platform for manufacturers and industrial suppliers. It connects shipment inspection, predefined quality requirements, and financial adjustments in one workflow.

AI-powered inspection: Analyzes delivery photographs to identify potential damage.
Agreement-based calculations: Applies predefined compensation rules to generate explainable settlement proposals.
Buyer-supplier collaboration: Allows both parties to review, accept, or dispute proposed adjustments.
Blockchain-backed verification: Records cryptographic evidence hashes and important decisions to create a tamper-evident audit trail. AI assists with evidence analysis, but it does not independently determine financial liability. Ambiguous cases can be flagged for human review.

How we built it

We designed ProofSettle around a transaction workflow that connects agreement creation, shipment evidence, inspection, settlement calculations, and dispute resolution. Our planned technology stack includes Python and FastAPI for backend APIs, Supabase for structured data and authentication, a pretrained vision model for damage analysis, and Solidity smart contracts for recording verifiable transaction events. The frontend uses React and TypeScript. Shipment documents and photographs remain off-chain, while cryptographic hashes can be recorded on-chain to help verify that evidence has not been silently modified. Challenges we faced

One of our biggest challenges was connecting AI-generated findings to financial decisions without blindly trusting model predictions. A photograph may show damage, but it cannot necessarily establish when the damage occurred or who was responsible. We addressed this in our design by separating AI-assisted inspection from deterministic settlement calculations, requiring predefined agreement rules, and preserving the ability for either party to dispute a result. Another challenge was deciding where blockchain adds genuine value. Rather than putting sensitive documents on-chain or using blockchain simply for payments, we focused on verifying evidence integrity and recording important decisions. What we learned

ProofSettle reinforced the importance of designing AI systems around accountability, not just automation. Reliable business decisions require transparent rules, accessible evidence, and a way for people to challenge uncertain results. Our goal is to make shipment disputes easier to understand, faster to review, and more transparent for everyone involved.