Your Second Brain. From Research to Execution.
Desktop-native AI workspace combining RAG Research, Project Management, and Content Creation
🚀 Quick Start · 📸 Screenshots · 📖 Documentation · 🤝 Contributing
- ✨ Introduction
- 🚀 Key Features
- 📸 Screenshots
- 🛠️ Tech Stack
- 🏗️ Getting Started
- 📚 User Guide
- 🗺️ Roadmap
- 🤝 Contributing
- 📄 License
NexusAI is a next-generation desktop-native AI workspace that transforms how you research, plan, and create. Built on .NET 8 LTS with Clean Architecture, it combines three powerful capabilities into one seamless experience:
📚 RAG Research → Chat with your documents (PDF, DOCX, PPTX, EPUB)
📊 Project Management → AI-powered Kanban boards with GitHub integration
🎨 Content Creation → Auto-generate presentations, wikis, and artifacts
|
Switch between two distinct AI personalities:
UI instantly transforms (colors, labels, tone) on toggle! |
Cloud Power + Local Privacy ☁️ Gemini 2.0 Flash:
- Multimodal (text + images)
- RAG with citations
- 15 req/min free tier
🔐 Ollama (Local):
- 100% private
- No internet required
- Models: llama3, mistral, etc.Switch providers at runtime, no restart needed! |
|
🌑 Void Background Almost pure black ( |
⚡ Neon Plasma Accent Electric gradient ( |
🪟 Glassmorphism Frosted surfaces, super-rounded corners (16-32px), and subtle glows create tactile depth. |
Zero instant changes — Every element animates in (300ms slide-up), buttons glow on hover (150ms), mode switches cross-fade (500ms).
| Feature | Description |
|---|---|
| 📚 RAG 2.0 Pipeline | Chat with PDF, DOCX, PPTX, EPUB, TXT, MD files using Retrieval-Augmented Generation |
| 🔍 Source Citations | Every AI response includes [filename.pdf] citations linking to exact sources |
| 🖼️ Multimodal Vision | Upload images for Gemini Vision analysis (diagrams, charts, screenshots) |
| 🎙️ Text-to-Speech | Listen to AI responses with integrated audio player (pause/resume) |
| 🌐 Multi-Document Context | Merge context from multiple files for comprehensive answers |
| ⚡ Streaming Responses | Real-time token streaming with progress indicators |
🔧 How RAG Works
┌──────────────────────────────────────────────────┐
│ 1. DOCUMENT INGESTION │
│ User uploads PDF/DOCX → Parser extracts text │
│ → Store in SessionContext.SourceDocuments │
└──────────────────────────────────────────────────┘
↓
┌──────────────────────────────────────────────────┐
│ 2. CONTEXT BUILDING │
│ Merge top 3 relevant docs into prompt │
│ Format: <Source>...[filename]...</Source> │
└──────────────────────────────────────────────────┘
↓
┌──────────────────────────────────────────────────┐
│ 3. AI GENERATION │
│ Gemini 2.0 Flash generates response │
│ → Extract citations via regex: \[.*?\] │
│ → Link citations to source documents │
└──────────────────────────────────────────────────┘
Tech Stack:
- PDF: iText7 (v8.0.5)
- Office: DocumentFormat.OpenXml (v3.2.0)
- eBooks: VersOne.Epub (v3.3.5)
- AI: Gemini 2.0 Flash, Ollama
NexusAI doesn't just chat — it creates real artifacts:
Input: "Create a REST API for blog"
Output:
- BlogController.cs
- BlogService.cs
- BlogRepository.cs
- Blog.cs (entity)
- Folder structure
Action: Download as .zip |
Input: Topic + slide count
Output:
- Title slide
- Content slides
- Speaker notes
- .pptx file
Action: Download + edit in PowerPoint |
Input: Documents + topic
Output:
- Hierarchical structure
- Chapter breakdown
- Tagged pages
- Markdown files
Action: Export to Obsidian vault |
Other Artifacts:
- 📖 Deep Dive: Executive summary + insights (Markdown)
- ❓ FAQ: 10-15 Q&A pairs from sources
- 📝 Study Guide: Learning objectives + questions
- 🎙️ Podcast Script: 2-person dialogue format
Example: |
Kanban Features:
|
Category Filtering:
- 💼 Work: Business projects
- 🎓 Education: Academic subjects
- 🏠 Personal: Hobby projects
Mode-aware filtering: Professional mode defaults to Work, Student mode to Education.
🎭 2026 Design Language (Click to expand)
/* Base */
--void-background: #050505; /* Almost pure black */
--obsidian-glass: #CC141416; /* 80% opacity dark grey */
/* Accent: Neon Plasma Gradient */
--deep-indigo: #6200EA;
--electric-violet: #B500FF;
/* Text Hierarchy */
--text-primary: #FFFFFF; /* Pure white, Bold */
--text-secondary: #B0B0C0; /* Cool grey */
--text-tertiary: #606070; /* Subtle labels */
/* Effects */
--frosted-edge: #33FFFFFF; /* Subtle inner glow */| Element | CornerRadius | Effect |
|---|---|---|
| Windows/Modals | 32px |
Super-rounded |
| Cards | 24px |
Obsidian Glass |
| Buttons | 16px |
Pill shapes |
| Inputs | 28px |
Floating capsules |
Since native BackdropFilter blur isn't available in WPF, we simulate glass via:
- Semi-transparent backgrounds:
#CC141416(80% opacity) - Colored drop shadows: Purple glow for elevation
- Frosted edge borders:
#33FFFFFF1px inner glow - Layered transparency: Multiple overlapping surfaces
Message Entrance:
- Slide up: TranslateY +20 → 0
- Fade in: Opacity 0 → 1
- Duration: 300ms
- Easing: CubicEase.EaseOut
Mode Switching:
- Cross-fade colors
- Duration: 500ms
- Easing: CubicEase.EaseInOut
Hover States:
- Scale: 1.0 → 1.05
- Glow: 0 → 24px blur
- Duration: 150ms
Card Lift (Drag):
- Opacity: 1.0 → 0.85
- TranslateY: 0 → -8px
- Purple shadow intensifies
- Duration: 150msZone 1: Window Chrome
- Custom draggable title bar (no standard Windows chrome)
- Obsidian Glass background blends into sidebar
- Logo uses Neon Plasma gradient
- Window controls (minimize/maximize/close) use
ChromeButtonstyle
Zone 2: Chat Interface
- User bubbles: Liquid Plasma gradient (RadialGradientBrush animated)
- AI bubbles: Obsidian Glass with white rim lighting
- Input field: Floating glass capsule with focus glow
- Messages slide up + fade in (300ms)
Zone 3: Kanban Board
- Columns have no visible borders (floating headers)
- Task cards: Obsidian Glass pills with
CornerRadius="24" - Drag effect: Card lifts (-8px), becomes translucent, emits purple glow
- All text uses TextPrimary/TextSecondary colors
No restart required! UI updates instantly via |
Coverage:
- UI labels, buttons, placeholders
- Error messages
- Settings panel
- Context menus
- Tooltips
Files:
- Resources/Languages/en-US.xaml
- Resources/Languages/ru-RU.xaml |
Persistent Preference: Language choice saved to settings.json and restored on app launch.
|
Concise AI responses + Project Kanban + Executive Purple theme |
Explanatory AI + Study Planner + Educational Orange/Teal theme |
|
Floating glass cards on Void background with Neon Plasma accents |
User bubbles with animated gradient + AI glass bubbles |
|
Interactive document relationship visualization |
AI-generated .pptx slides with speaker notes |
Runtime: .NET 8.0 LTS (Long-Term Support)
Language: C# 12 (Primary Constructors, Collection Expressions, File-Scoped Namespaces)
UI Framework: WPF (Windows Presentation Foundation)
Database: SQLite + Entity Framework Core 8.0
Architecture: Clean Architecture (4 layers: Domain → Application → Infrastructure → Presentation)
Patterns: MVVM, CQRS, Repository, Factory, Strategy, Railway Oriented Programming| Provider | Model | Use Case | Features |
|---|---|---|---|
| 🌐 Google Gemini | gemini-2.0-flash |
Cloud AI |
✅ Multimodal (text + images) ✅ Strict RAG with citations ✅ 15 req/min free tier ✅ Streaming responses |
| 🔐 Ollama | llama3, mistral, etc. |
Local LLM |
✅ 100% offline ✅ Privacy-first ✅ No API costs ✅ Custom models |
PDF: iText7 (v8.0.5)
Office: DocumentFormat.OpenXml (v3.2.0) - DOCX, PPTX
eBooks: VersOne.Epub (v3.3.5)
Text: Native support for TXT, MDTheme: MaterialDesignInXamlToolkit (v5.1.0)
Icons: Material Design Icons (2,000+ icons)
Design System: Custom Cyber-Noir Glass (Resources/Styles/DarkNeuralGlass.xaml)
Animations: WPF Storyboards, DoubleAnimation, ColorAnimation
MVVM: CommunityToolkit.Mvvm (v8.4.0) - ObservableProperty, RelayCommand🏛️ Clean Architecture (Click to expand)
┌──────────────────────────────────────────────────────────────┐
│ NexusAI.Presentation │ ← User Interface
│ ┌────────────────────────────────────────────────────────┐ │
│ │ • ViewModels (MVVM) │ │
│ │ • Views (XAML) │ │
│ │ • Converters (BoolToVisibility, ModeToColor) │ │
│ │ • Resources (DarkNeuralGlass.xaml, Languages) │ │
│ │ Technology: WPF, MaterialDesign, CommunityToolkit │ │
│ └────────────────────────────────────────────────────────┘ │
└──────────────────────────┬───────────────────────────────────┘
│ depends on ↓
┌──────────────────────────▼───────────────────────────────────┐
│ NexusAI.Application │ ← Business Logic
│ ┌────────────────────────────────────────────────────────┐ │
│ │ • Use Cases (Commands/Queries with Handlers) │ │
│ │ • Interfaces (IAiService, IProjectService) │ │
│ │ • Services (SessionContext, KnowledgeGraphService) │ │
│ │ • DTOs (Data Transfer Objects) │ │
│ │ Pattern: CQRS, Result<T> (Railway Oriented) │ │
│ └────────────────────────────────────────────────────────┘ │
└──────────────────────────┬───────────────────────────────────┘
│ depends on ↓
┌──────────────────────────▼───────────────────────────────────┐
│ NexusAI.Domain │ ← Core (Pure C#)
│ ┌────────────────────────────────────────────────────────┐ │
│ │ • Entities (User, Project, ProjectTask, ChatMessage) │ │
│ │ • Value Objects │ │
│ │ • Enums (AppMode, AiProvider, TaskStatus) │ │
│ │ • Result<T> (Railway Oriented Programming) │ │
│ │ ✅ ZERO external dependencies │ │
│ └────────────────────────────────────────────────────────┘ │
└──────────────────────────▲───────────────────────────────────┘
│ depends on ↑
┌──────────────────────────┴───────────────────────────────────┐
│ NexusAI.Infrastructure │ ← External Concerns
│ ┌────────────────────────────────────────────────────────┐ │
│ │ • Persistence (AppDbContext, EF Core migrations) │ │
│ │ • AI Services (GeminiAiService, OllamaService) │ │
│ │ • Parsers (PdfParser, WordParser, EpubParser) │ │
│ │ • External APIs (HTTP clients, Obsidian sync) │ │
│ │ Technology: EF Core, HttpClient, iText7, OpenXml │ │
│ └────────────────────────────────────────────────────────┘ │
└──────────────────────────────────────────────────────────────┘
Dependency Rule: All dependencies point INWARD toward Domain.
+ ✅ Presentation → Application → Domain
+ ✅ Infrastructure → Application → Domain
+ ❌ Domain → (NO dependencies on outer layers)Design Patterns:
| Pattern | Usage | Location |
|---|---|---|
| MVVM | UI separation | Presentation/ViewModels/ |
| CQRS | Commands/Queries | Application/UseCases/ |
| Repository | Data access abstraction | IProjectService, IAuthService |
| Factory | Runtime provider switching | IAiServiceFactory, IDocumentParserFactory |
| Strategy | Parser selection | IDocumentParser (PDF/Word/EPUB) |
| Railway Oriented | Error handling | Result<T> + extensions |
| Primary Constructor | DI injection (C# 12) | All services |
OS: Windows 10/11 (x64)
Runtime: .NET 8.0 SDK
RAM: 4 GB minimum, 8 GB recommended
Storage: 500 MB for app + documents
Display: 1920x1080 or higher (for optimal glass UI)- Go to Releases
- Download
NexusAI-v1.0.0-win-x64.zip - Extract to a folder (e.g.,
C:\Apps\NexusAI) - Run
NexusAI.exe - Enter your Gemini API key (get one at aistudio.google.com)
# 1. Clone repository
git clone https://github.com/yourusername/NexusAI.git
cd NexusAI
# 2. Restore dependencies
dotnet restore
# 3. Build solution (Debug or Release)
dotnet build --configuration Release
# 4. Run application
cd src/NexusAI.Presentation
dotnet run
# Or open NexusAI.sln in Visual Studio 2022 and press F5- Visit Google AI Studio
- Create a free API key (15 requests/minute)
- Paste into the API Key field in the app header
# 1. Download Ollama from https://ollama.ai/
# 2. Install and start the service
# 3. Pull a model (e.g., llama3)
ollama pull llama3
# 4. In NexusAI, select "Ollama" as AI provider
# 5. Choose your downloaded model from dropdownNo API key needed for Ollama — 100% offline!
|
📁 File Dialog
|
🖱️ Drag & Drop
|
📝 Obsidian Vault
|
Document Management:
- ✅ Toggle checkbox to include/exclude from AI context
- ✅ Click ✕ to remove document
- ✅ Hover to see file size and load timestamp
💼 Professional Mode Example:
You: "Analyze Q4 revenue trends from the financial report."
AI: "Revenue increased 23% YoY to $4.2M [Q4_Report.pdf].
Key drivers: Enterprise segment (+35%), Product A sales..."
🎓 Student Mode Example:
You: "What is photosynthesis?"
AI: "Great question! Let's break it down step-by-step.
Think of a plant as a tiny solar panel factory [biology_ch3.pdf].
The process has two main stages:
1. Light-dependent reactions (capturing energy)
2. Light-independent reactions (making glucose)
Let me explain each..."Citations: All responses include [filename.pdf] links. Click to highlight source in sidebar.
Toggle between modes instantly (no restart):
| Feature | 💼 Professional | 🎓 Student |
|---|---|---|
| AI Tone | Concise, business-focused | Explanatory, teaching |
| UI Labels | "Projects", "Tasks" | "Subjects", "Assignments" |
| Accent Color | Deep Purple (#6200EA) |
Orange/Teal |
| Default Category | Work | Education |
Toggle Location: Top header (💼 ⇄ 🎓 icons)
1. Go to "Projects" tab
2. Click "➕ NEW PROJECT"
3. Fill in:
- Title: "Website Redesign"
- Description: "Modern UI overhaul with dark mode"
- GitHub Repo: https://github.com/company/website (optional)
- Category: Work / Education / Personal
4. Click "Generate Plan" → AI creates task breakdown
Drag-and-drop:
- Cards automatically update
OrderIndexin database - Smooth animations (lift effect on grab)
- Purple glow when hovering over drop zone
| Type | Output | Use Case |
|---|---|---|
| 📖 Deep Dive | Markdown | Executive summary + insights |
| 📄 Summary | 2-4 paragraphs | Quick overview |
| 📚 Study Guide | Structured outline | Learning objectives + questions |
| ❓ FAQ | 10-15 Q&A | Common questions from sources |
| 🎙️ Podcast Script | Dialogue | 2-person conversation |
1. Go to "Presentation" tab
2. Enter:
- Topic: "AI in Healthcare"
- Slide count: 10
3. Click "Generate Deck"
4. AI creates:
- Title slide
- Content slides (bullets)
- Speaker notes
- Conclusion
5. Download as .pptx
6. Edit in PowerPoint- 🌐 Language: English / Russian (instant switch)
- 🤖 AI Provider: Gemini / Ollama
- 🔑 API Key: Paste Gemini key
- 🔊 Audio: Volume, voice, speed
✅ Completed (v1.0)
- Gemini 2.0 Flash integration with streaming
- Ollama local LLM support
- RAG pipeline with source citations
- PDF, DOCX, PPTX, EPUB parsing
- Dual mode (Professional/Student)
- Cyber-Noir Glass UI (2026 design)
- Kanban board with drag-and-drop
- AI project planning
- PowerPoint generation
- Wiki system
- Knowledge graph
- Obsidian integration
- Code scaffolding
- Text-to-Speech
- Localization (English/Russian)
- Clean Architecture implementation
- SQLite + EF Core persistence
-
Performance
- Lazy loading for document lists (1000+ files)
- Virtual scrolling in chat
- Background indexing for RAG
- Database connection pooling optimization
-
UX Enhancements
- Auto-scroll to latest message
- Ctrl+F search within documents
- Document preview modal
- Keyboard shortcuts (Ctrl+K command palette)
- Undo/redo for task moves
-
Data Portability
- Export chat history to PDF/Markdown
- Import/export projects as JSON
- Backup/restore database
- Bulk document import folder
-
AI Providers
- OpenAI GPT-4 integration
- Anthropic Claude support
- Azure OpenAI Service
- Custom API endpoint configuration
-
Advanced Features
- Vector embeddings (ChromaDB/Qdrant)
- Semantic search across documents
- Multi-user collaboration (SQLite → PostgreSQL)
- Cloud sync (optional, Azure/AWS)
- Voice input (speech-to-text)
-
Integrations
- Notion sync
- Confluence integration
- Google Drive import
- Slack notifications
-
Enterprise
- SSO/SAML authentication
- Role-based access control (RBAC)
- Audit logging
- Self-hosted server option
-
AI Enhancements
- Fine-tuned models (custom RAG)
- Prompt template library
- Multi-agent workflows
- Autonomous task execution
-
Platform Expansion
- Web version (Blazor)
- Mobile app (MAUI)
- Linux support (Avalonia UI)
Contributions are highly welcome! Whether it's bug fixes, new features, or documentation improvements.
# 1. Fork repository
gh repo fork yourusername/NexusAI
# 2. Create feature branch
git checkout -b feature/amazing-feature
# 3. Make changes (follow C# 12 style)
# 4. Commit with conventional commits
git commit -m "feat: add export to JSON"
# 5. Push and open PR
git push origin feature/amazing-feature📝 Code Style (Click to expand)
✅ DO:
- Use file-scoped namespaces (namespace X;)
- Use primary constructors for DI (public class Service(IRepo repo))
- Use collection expressions (List<int> x = [];)
- Use guard clauses (early returns)
- Return Result<T> from business logic
- Add XML documentation for public APIs
❌ DON'T:
- Throw exceptions in Domain/Application layers
- Reference Infrastructure from Application layer
- Use magic strings (use constants/resources)
- Add "AI comments" that explain what code does
- Nest logic more than 3 levels deepExample:
// ✅ GOOD: Primary constructor, guard clauses, Result<T>
public sealed class ProjectService(AppDbContext context) : IProjectService
{
public async Task<Result<Project>> CreateAsync(string title, ...)
{
// Guard clause (early return)
if (string.IsNullOrWhiteSpace(title))
return Result<Project>.Failure("Title cannot be empty");
var project = new Project { Title = title, ... };
context.Projects.Add(project);
await context.SaveChangesAsync();
return Result<Project>.Success(project);
}
}🎯 Commit Message Format
feat: New feature (e.g., "feat: add Notion integration")
fix: Bug fix (e.g., "fix: resolve null ref in ProjectService")
docs: Documentation (e.g., "docs: update README with Docker setup")
style: Code style/formatting (no logic change)
refactor: Code restructuring (no behavior change)
perf: Performance improvement
test: Adding/updating tests
chore: Build process, dependencies
- 🧪 Testing: Unit tests, integration tests (xUnit, FluentAssertions)
- 🌐 Localization: German, Spanish, French translations
- 🎨 UI/UX: Design improvements, accessibility (WCAG)
- 📚 Documentation: Tutorials, video guides, API docs
- 🐛 Bug Fixes: See Issues
Before contributing, please read CONTRIBUTING.md for detailed guidelines.
This project is licensed under the MIT License.
MIT License
Copyright (c) 2026 NexusAI Contributors
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
TL;DR: Free to use, modify, and distribute. Just include the copyright notice.
Built with ❤️ using these amazing open-source projects:
|
AI & ML
|
UI & Design
|
|
Document Processing
|
Data & Infrastructure
|
- 📖 Documentation: Wiki
- 💬 Discussions: GitHub Discussions
- 🐛 Bug Reports: Issues
- 📧 Contact: vitalydoxr@gmail.com
- ⭐ Star this repo to receive notifications
- 👀 Watch releases for new versions
- 🔔 Subscribe to Discussions for announcements
Built with ❤️ using .NET 8 LTS and Modern C# 12
NexusAI — Your Intelligent Workspace Companion
v1.0.0 • Download • Documentation • Report Bug • Request Feature
Made with 🧠 by developers, for developers
© 2026 NexusAI Contributors • MIT License