Inspiration
What it does
Inspiration
Field-service contractors often have job photos, work orders, invoices, and customer sign-offs scattered across folders. Finding evidence of completed work can take hours.
What it does
ProofPath is an AI Skill prototype for UGREEN AI NAS designed to organize authorized job files into traceable evidence reports. It identifies available photos and documents, highlights missing records, and flags uncertain matches for human review.
How we built it
We created an original Uliya-format SKILL.md workflow and supporting evidence rules using AI-assisted development with ChatGPT.
The design uses authorized UGREEN file, document, and knowledge capabilities. It emphasizes local data storage, restricted permissions, and human approval before saving reports.
Challenges
The main challenge is preventing AI from inventing evidence or incorrectly matching documents to jobs. ProofPath addresses this through source references, confidence classifications, and explicit missing-evidence reporting.
Accomplishments
Completed the initial Skill definition, evidence-classification rules, and planned testing scenarios.
What's next
Validate the Skill with the official UGOS creation tool, test it on a UGREEN AI NAS if selected as a finalist, and refine the workflow based on actual test results.
Development status: Initial prototype documentation complete. NAS runtime testing and official creation-tool validation remain pending.
How we built it
Challenges we ran into
Accomplishments that we're proud of
What we learned
What's next for ProofPath
Built With
- ai-agents
- markdown
- ugos
- uliya
