Inspiration
A purchase order rarely tells the whole story. Supplier confirmations, delivery notes, and packing slips often arrive as separate documents. When quantities change or shipments are delayed, important discrepancies can disappear across folders and email attachments.
EvidenceChain NAS is designed to make these hidden exceptions visible, traceable, and actionable.
What it does
EvidenceChain NAS is an AI Skill for UGREEN AI NAS that aims to cross-check the evidence behind a supplier delivery.
Rather than simply summarizing documents, it identifies potential inconsistencies between purchase orders, supplier confirmations, and packing slips.
Each finding is designed to include its supporting source, missing information, and a suggested verification question.
For example:
- Purchase order: 100 units ordered.
- Supplier confirmation: 80 available, 20 backordered.
- Packing slip: 75 units listed.
- Expected review: a possible 5-unit delivery gap and a separate 20-unit backorder, pending verification.
The goal is to prevent unresolved issues from being overlooked while keeping people in control of decisions.
How it works
The proposed Skill uses the official UGREEN Skill Developer Package and Uliya Skill Development Standard.
- The user selects an authorized NAS folder and provides an order reference.
- The file agent locates relevant documents through supported filesystem capabilities.
- The document agent extracts purchase-order information and compares available evidence.
- EvidenceChain organizes potential discrepancies, source references, uncertainties, and missing documents into a review report.
- The user reviews the findings before taking further action.
The design uses registered expert agents, respects access permissions, and does not automatically change accounting or inventory records.
What makes it different
EvidenceChain focuses on supplier fulfillment evidence, rather than generic document summaries or invoice extraction.
Its proposed differentiators are:
- Cross-document exception detection.
- Separate treatment of delivery shortages and backorders.
- Source-linked findings and explicit uncertainty.
- Human verification before operational decisions.
- A workflow centered on authorized NAS documents.
Current development status
The concept and official developer capabilities have been reviewed. The initial Uliya Skill package is being prepared.
The workflow has not yet been validated on a physical UGREEN AI NAS. Runtime testing, extraction accuracy, error handling, and final reporting behavior remain to be verified.
What's next
If selected as a finalist, the next step is to test the Skill on the assigned UGREEN AI NAS environment using synthetic supplier documents, verify its limitations, and refine the workflow.
The longer-term vision is a dependable document intelligence capability that helps small businesses detect operational exceptions before they become costly problems.
Built With
- agents
- ai
- markdown
- skill
- ug-document
- ug-filesystem
- ugos
- ugreen
- uliya
