-
From conversations to everyday workflows, EVIE aims to bring personal productivity and intelligent assistance together.
-
EVIE’s native macOS application brings conversational AI and personal assistance into a dedicated desktop experience.
-
EVIE’s conversational AI foundation enables intelligent interactions through an actively developed application interface.
-
One AI companion, multiple platforms. EVIE is evolving toward connected experiences across desktop, web, and mobile.
-
Proposed UGREEN AI NAS Skill: natural-language file discovery, personal knowledge, smart organization, and privacy-aware automation.
-
EVIE’s evolving AI ecosystem explores multiple intelligence providers and personalized assistance within one unified experience.
-
EVIE Admin provides a foundation for authorized administration, access controls, and future ecosystem management.
-
EVIE’s web experience extends its conversational interface into the browser, building toward a unified cross-platform ecosystem.
-
Privacy-aware architecture, controlled permissions, and secure access are central to EVIE’s ongoing development.
-
A unified EVIE Account experience connects identity, preferences, and access across the evolving product ecosystem.
-
EVIE’s extensible architecture is designed to support plugins, connected services, and new AI-powered capabilities.
-
EVIE’s connected ecosystem expands toward UGREEN AI NAS with a proposed Skill for intelligent, privacy-aware personal storage.
-
More than a chatbot. EVIE combines a distinctive AI identity with personalized conversations and an evolving digital experience.
-
EVIE’s journey continues beyond its existing prototype, connecting personal AI, intelligent automation, and private data.
-
Meet EVIE, a personal AI companion already in active development, with an existing prototype and a growing ecosystem.
Inspiration
Artificial intelligence is becoming increasingly powerful, yet our digital lives remain fragmented across applications, devices, cloud services, and personal storage systems.
I started building EVIE with a different vision: one personal AI companion capable of bringing intelligence, connected services, and everyday assistance together through natural conversation.
I believe people should be able to interact with their digital environments without constantly switching between applications or learning complicated interfaces.
EVIE is an independent software project that predates this Hackathon. It already has an evolving application prototype, AI integration, account infrastructure, and ongoing development across multiple platforms.
The UGREEN AI NAS Hackathon presents an exciting opportunity to extend that vision into personal storage.
What if your NAS became more than a place to store files? What if you could simply talk to your digital world?
That question inspired EVIE’s new UGREEN Skill.
What it does
EVIE — Universal NAS AI Companion
EVIE’s long-term vision is to become a unified conversational companion for the digital content and services available through a personal NAS.
Instead of navigating multiple applications for documents, photos, movies, music, and other personal information, users could interact with supported NAS capabilities through natural-language requests.
The envisioned experience spans several areas:
Personal Files & Documents
- Discover files through natural-language requests.
- Ask questions about documents and receive evidence-backed answers.
- Summarize information and identify relevant sources.
- Explore safe file organization workflows.
Photos & Personal Memories
- Discover photos and videos using supported metadata.
- Explore personal collections and albums.
- Connect memories with relevant personal information.
Movies & Entertainment
- Discover locally stored movies and videos.
- Explore available media metadata.
- Compare media files and access supported playback capabilities.
Music
- Search personal music collections.
- Discover artists, albums, and tracks.
- Interact with supported music capabilities.
Recordings & Personal Knowledge
- Explore recorded information through supported transcription capabilities.
- Compare information from different sources.
- Generate structured summaries and decision briefs.
Intelligent Workflows
- Combine supported NAS operations into useful multi-step experiences.
- Respect user permissions and confirmation requirements.
- Prioritize transparency, privacy, and user control.
These areas represent the broader product vision. Their availability depends on the documented UGREEN interfaces, permissions, and successful runtime validation.
The Hackathon Skill: EVIE Personal NAS Intelligence
For Round 1, I developed a new standalone Uliya Skill focused on a specific, practical workflow:
Discover → Select → Understand → Answer
A user can request a document from an authorized NAS folder, select the correct matching PDF, and ask a question or request a summary.
The Skill coordinates supported UGREEN filesystem and document experts to produce an evidence-backed response.
The current scope is deliberately constrained to one authorized folder and one selected text-based PDF at a time.
This foundation demonstrates the interaction model behind the larger Universal NAS AI Companion vision.
The Next Experience: Cross-Source Intelligence
If selected as a finalist, the proposed next step is a Project Decision Brief.
Imagine having a project requirements document and a recording of a meeting where important decisions were made.
The envisioned workflow would allow a user to ask:
“EVIE, compare the original requirements with the decisions from our latest meeting. What changed, what was agreed, and what remains unresolved?”
EVIE would combine evidence from the selected document and recording to produce a structured comparison.
This extension depends on verifying the required transcription, permissions, file identity, local processing, and source-evidence capabilities on actual UGREEN NAS hardware.
It is a proposed finalist extension, not an implemented Round 1 feature.
How I built it
EVIE is developed through an AI-assisted software engineering workflow combining architecture, application development, backend integration, interface design, testing, and iterative improvement.
The existing EVIE ecosystem includes a Tauri-based macOS application, React and TypeScript interfaces, SQLite-backed conversation storage, OpenAI integration, and account infrastructure.
For this Hackathon, I created EVIE Personal NAS Intelligence as a separate UGREEN-compatible Skill rather than modifying the existing EVIE application.
The Skill follows the official Uliya Skill Development Standard and uses documented UGREEN expert capabilities.
Its workflow includes:
- Interpret a user’s document discovery request.
- Search within a specified authorized NAS folder.
- Return matching PDF candidates.
- Require explicit selection when multiple files match.
- Delegate the selected document to the document expert.
- Return an answer or summary with supporting evidence.
- Stop safely when access, source identity, or required processing conditions cannot be verified.
The Skill uses documented filesystem and document actions and is designed around read-only operations.
Its authoring and validation used the official UGREEN development toolkit and OpenAI Codex.
The completed Round 1 package passed the official static checker with zero errors and zero warnings.
This confirms static compliance checks, not successful execution on physical NAS hardware.
Challenges I ran into
One of the biggest challenges was balancing EVIE’s ambitious product vision with the actual technical capabilities exposed through the UGREEN Skill interface.
The official capability audit identified 78 documented Actions across nine CLI families.
Of these, 15 were classified as confirmed at the documentation level, while 63 require additional conditions, permissions, dependencies, or runtime verification.
This made it important to distinguish between what is documented, what is technically conditional, and what remains outside the currently audited interface.
Other important challenges included:
- Maintaining strict user authorization boundaries.
- Preventing unauthorized cross-folder access.
- Safely passing file identities between different experts.
- Avoiding unsupported assumptions about NAS APIs.
- Handling incomplete or unavailable source evidence.
- Designing workflows without access to finalist hardware.
- Ensuring that the new Hackathon Skill remains independent from the pre-existing EVIE application.
These constraints helped shape a more focused, secure, and technically credible first implementation.
Accomplishments that I’m proud of
EVIE has progressed beyond an initial concept into an actively developed software ecosystem.
The broader project already includes application infrastructure, AI integration, native desktop development, browser interfaces, and a recognizable product identity.
For the UGREEN Hackathon, I am particularly proud of having:
- Created a new standalone Uliya Skill.
- Designed a permission-aware document discovery and question-answering workflow.
- Integrated documented filesystem and document expert delegation.
- Established explicit handling for ambiguous search results and incomplete evidence.
- Completed a technical audit of the available UGREEN Skill capabilities.
- Defined a feasible path toward cross-source AI workflows.
- Passed the official Skill static checker with zero errors and warnings.
- Prepared the required technical documentation and validation materials.
Most importantly, the Hackathon work establishes a concrete technical starting point for EVIE’s broader NAS integration vision.
What I learned
This project reinforced an important principle:
A powerful AI companion is not defined by how many features it promises, but by how reliably it connects useful capabilities.
Working with the UGREEN development environment highlighted the importance of documented interfaces, explicit permissions, trustworthy source evidence, and safe orchestration.
It also demonstrated how a focused Skill can become the foundation for more sophisticated experiences.
AI-assisted development accelerated research, documentation, and implementation, but careful verification and architectural decisions remained essential.
The experience strengthened my belief that the future of personal AI lies in useful, secure interactions with the information and services people already own.
What’s next for EVIE
Round 1 — Personal NAS Intelligence
The initial Skill package and technical proposal have been prepared and statically validated.
The current workflow focuses on authorized PDF discovery and evidence-backed document understanding.
Finalist Stage — Cross-Source Intelligence
If selected as a finalist, the first priority will be validating the existing Skill on actual UGREEN NAS hardware.
Testing will focus on installation, expert behavior, permissions, file identity, local processing, and answer quality.
If those checks succeed, the next objective will be exploring the Project Decision Brief workflow, combining a selected document with a recording through supported NAS capabilities.
Long-Term Vision — Universal NAS AI Companion
Beyond the Hackathon, EVIE aims to evolve toward a broader conversational experience across personal files, documents, photos, videos, music, and other supported NAS services.
Future capabilities will depend on available platform interfaces, security requirements, and implementation testing.
The goal is not to replace the NAS operating system or reproduce its applications.
The goal is to make the capabilities of a personal NAS work together more naturally through one intelligent companion.
One AI companion. Your entire digital world. One conversation.
EVIE is already being built.
The UGREEN Hackathon is an opportunity to bring that vision closer to the information people own and control.
Built With
- ai-agents
- artificial-intelligence
- css
- git
- github
- html
- ios
- javascript
- llm
- macos
- nas
- node.js
- react
- rest-api
- rust
- tauri
- typescript
- ugreen
