InspirationPROOFLOOP: Turning Contractual Promises into Verifiable Proof
Inspiration
Businesses sign contracts with suppliers and service providers every day. Those contracts contain promises: inspections must happen, equipment must be maintained, certificates must remain valid, and services must be delivered on time.
But the evidence is often scattered across folders containing invoices, inspection reports, photographs, work orders and certificates.
We were inspired by a simple question:
What if a business could verify whether contractual promises are actually supported by the evidence stored in its own NAS?
That question led to PROOFLOOP.
What PROOFLOOP Does
PROOFLOOP is an AI-powered contract-to-evidence verification Skill designed for UGREEN AI NAS.
Instead of simply summarizing contracts, it connects individual contractual obligations to relevant operational records.
It aims to:
- Identify obligations, deadlines and conditions in contracts.
- Find relevant evidence across authorized NAS files.
- Compare contractual requirements against available records.
- Identify missing evidence, expired documents and conflicting information.
- Produce structured findings linked to their source documents.
Its central concept is an Obligation-to-Evidence Graph: a traceable relationship between what was promised and the records available to support it.
PROOFLOOP distinguishes between supported findings, unverified obligations and issues requiring human review. It does not make legal judgments or automatically declare contractual breaches.
How We Built It
We developed the initial Skill using the official UGREEN Skill Creation Toolkit and its documented capabilities.
The design organizes the workflow into three specialist responsibilities:
- Obligation Analyst: Identifies promises and requirements in contracts.
- Evidence Investigator: Locates potentially relevant records.
- Verification Analyst: Compares evidence with obligations and prepares traceable findings.
The workflow is designed around authorized NAS access, source references and human oversight.
The initial submission package contains the Skill instructions and supporting Markdown documentation.
What We Learned
We learned that extracting information from a contract is only the beginning. The more valuable challenge is connecting that information to real operational evidence.
We also learned the importance of designing AI workflows that distinguish verified information from assumptions, preserve source references and leave consequential decisions to humans.
Challenges
The biggest challenge was working within the UGREEN Skill environment and ensuring the design followed its supported capabilities rather than assuming unrestricted code execution.
Another challenge was making the project meaningfully different from conventional AI contract summarizers and contract-management tools.
We addressed this by centering the project on evidence relationships, missing-proof detection and traceable verification.
What's Next
Our next objective is to test PROOFLOOP against realistic contract and evidence archives, improve its verification accuracy and validate the complete workflow on UGREEN AI NAS hardware.
PROOFLOOP: Because a promise should be supported by proof.
Log in or sign up for Devpost to join the conversation.