Projects

AI Council – Multi-Agent Discussion Platform
Type: Software Status: Alpha Tech Stack: Python, FastAPI, SQLAlchemy 2.0 (async), aiosqlite, SQLite, Vanilla JS, Tailwind CSS, WebSocket Problem Statement Individual AI models have inherent biases and blind spots. Users receive only one perspective on complex questions, depending on the chosen model and provider. There is no simple way to systematically have multiple AI models with different ‘personalities’ debate each other and consolidate through voting. Cross-provider comparisons require manual copy-paste between different interfaces. Description A web application where multiple AI agents with their own personalities and models discuss, debate, and
AI Writer – AI-Powered Blog Writing Assistant
Type: Software Status: In Development Tech Stack: Python 3.12, FastAPI, SQLAlchemy, SQLite, React 18, Vite 6, Tailwind CSS, OpenAI API, Anthropic API, Docker Problem Statement Existing AI writing tools either generate complete texts (which dilutes authorship) or offer only spell-checking (which provides too little support). A middle ground is missing: a tool that provides contextual suggestions paragraph by paragraph, which the author can individually accept or reject while maintaining creative control. Multilingual support with translation functionality is rarely integrated in existing tools. Description A writing assistant for blog posts that

Anonymizer
Type: Software Status: Alpha Tech Stack: Python, spaCy (NER), Faker, cryptography (Fernet/PBKDF2), Click, pathspec, chardet, tkinter, rich Problem Statement Developers and companies want to use AI tools for code reviews and debugging but cannot send code with real API keys, passwords, company names, or personal data to external services due to privacy and security concerns. Manual redaction is error-prone and time-consuming. No tool exists that automatically detects and reversibly anonymizes both technical secrets and contextual personal data. Description Anonymizes sensitive data in code and text files before sharing them with

AppForge – Low-Code Application Builder
Type: Software Status: Prototype Tech Stack: Python, FastAPI, SQLAlchemy, Alembic, Pydantic, JWT, React 18 (TypeScript), TanStack Query, Tailwind CSS, Vite, React Flow, PostgreSQL, Docker, Nginx Problem Statement Developing database-driven web applications requires deep knowledge of backend frameworks, ORM configuration, database migrations, API design, and frontend development. Existing low-code platforms are either proprietary and cloud-bound, offer no real code generation, or create vendor lock-in. A self-hosted solution that enables both visual design and standalone, deployable code is missing. Description A comprehensive low-code platform for visually building database-driven web applications. Offers five

ARACHNE – Autonomous Research Agent Cluster with Hierarchical Navigation Engine
Type: Software Status: POC Tech Stack: Python, FastAPI, Ollama, Playwright, ChromaDB (RAG), SQLite, React, WebSocket, Docker Problem Statement Manual research on complex, rapidly changing topics (crisis situations, markets, technology trends) is time-consuming, error-prone, and not scalable. Individual search engines only provide snapshots, not continuously updated, quality-assessed knowledge bases. Existing monitoring tools are limited to specific data sources and offer no intelligent filtering or deduplication. Description A distributed, AI-driven multi-agent research system that autonomously collects, evaluates, and builds structured knowledge from the web. A fast orchestrator model distributes research tasks to

Captioner – Image Captioning for AI Training
Type: Software Status: Beta Tech Stack: Python, FastAPI, Uvicorn, Transformers (Florence-2, JoyCaption), PyTorch, React, Vite, TypeScript Problem Statement Training AI image models (LoRA, DreamBooth, fine-tuning) requires image datasets with high-quality text descriptions. Manual captioning is extremely time-consuming, and automatic tools rarely offer a visual gallery view, multi-model support, or the ability to switch between caption formats (full text, tags, short description). Batch processing with subsequent manual correction in an integrated workflow is missing. Description A lean tool for creating image descriptions for AI training datasets. Images are displayed in a
Claudestro: ProjectMind – Semantic Project Memory
Type: Software Status: In Development Tech Stack: Python, SQLite (WAL), tree-sitter, FastMCP, Starlette, watchdog, AST-Modul Problem Statement AI code assistants like Claude Code lose their context about project structure with every restart. They don’t know which functions exist, who calls them, or what impact a change has. In large codebases, this leads to redundant analyses, false assumptions, and breakpoints. A persistent, semantic knowledge base that preserves code relationships across sessions and is available as an MCP server for AI assistants is missing. Description A persistent knowledge base for code projects

Dailys – AI News Project
About the project: The “Dailys” art-project is a combination of automated web scraping, AI text analysis, and generative AI art. Daily, news content is automatically collected from www.orf.at. Following this, a specialized AI utilizes this data to create a summary of the most important events. Based on this summary, the AI analyzes the prevailing mood of the news to construct a “mood picture of the day.” Based on this mood picture, the AI finally designs a visual artwork that captures the essential atmosphere of the day and presents it in the
DiffDirs – Recursive Directory Comparison
Type: Software Status: V.1.0 Tech Stack: Python, PyYAML, difflib, hashlib Problem Statement Many legacy projects, especially in government agencies and enterprises, do not use version control systems (Git). When code needs to be synchronized between development and production environments, a simple tool for comparing two directory structures is missing. Existing diff tools usually work at file level, not recursively, and offer no clear reports with statistics and filter options. Description A Python tool for recursive comparison of two project directories — designed for projects without Git. Detects new, deleted, and
Elementor-Manager
Type: Software Status: Prototype Tech Stack: Python, Flask, MySQL, SQLite, Jinja2, MCP (stdio/SSE), REST API Problem Statement Managing multiple WordPress websites with Elementor builder requires repetitive manual work across different admin interfaces. Content changes across multiple sites (e.g., contact details, YouTube embeds, footer texts) must be performed individually. Elementor stores its data in serialized JSON structures in the database, making direct manipulation without specialized tools practically impossible. Description A Python web application for direct management and automation of WordPress Elementor pages via MySQL database access. Offers multi-site management, page/post/menu administration,

Ghost in a Box
Ghost in a Box is an art project that specializes in connecting various creative disciplines, including digital art, AI art, photography, and blockchain technology. It creates impressive and thought-provoking artworks that harmoniously combine both analog and digital elements. In doing so, it bridges the gap between classical and modern art forms, deepening the understanding of the complex relationships between these seemingly different worlds. Ghost in a Box also utilizes blockchain technology to create digital artworks that are genuine, limited, and secure, positioning itself at the forefront of a new movement
godM0d3 – Modular Self-Hosted Plugin Platform
Type: Software Status: Beta Tech Stack: Python, FastAPI, SQLAlchemy 2 (async), Pydantic v2, Celery 5, Redis 7, React 18 (TypeScript), Vite 5, Tailwind CSS 3, PostgreSQL 16, Ollama, Docker Compose Problem Statement Self-hosted productivity and AI tools are fragmented: each service (chat, email, calendar, invoicing, knowledge base) runs as a separate application with its own login, database, and maintenance. There is no modular, extensible platform that unifies these services under one roof while ensuring true data isolation between plugins. Existing all-in-one solutions are either monolithic or cloud-based. Description A modular,

Hidden Galleries
Type: Software Status: Beta Tech Stack: PHP, JavaScript (V1) — Python, React, FastAPI, NFC, PostgreSQL, Leaflet (V2) Problem Statement Public art is typically large, obvious, and institutionally curated. No system connects miniature artworks in secret locations with a digital collecting and discovery mechanic. The connection between physical discovery (treasure hunt principle) and digital authentication and collection (NFC + blockchain) does not exist as a platform. Existing geocaching platforms have no art focus and no cryptographic authenticity verification. Description A platform for hidden miniature art installations in public spaces that connects

kAI zero
kAI zero is an experimental art and music project that explores the creative potential of artificial intelligence. At its core, the project is both an artistic experiment and a cultural inquiry: how far can machine-generated music go, and what happens when technology becomes a co-creator rather than just a tool? The debut album, released on August 22, 2025, features 25 original tracks that merge human ideas with AI-driven composition. Each piece reflects a different facet of this collaboration – from abstract soundscapes to rhythm-driven explorations – resulting in a body

kAI zero Player
Type: Software Status: Beta Tech Stack: Python, python-vlc, pygame (Joystick), tkinter Problem Statement Art installations and exhibitions often need a dedicated, robust video player that works on retro hardware (CRT TVs) with unconventional input devices (joysticks). Standard media players are oversized for these scenarios, offer no joystick control, and no CRT-optimized interface. A minimalist, exhibition-ready player is missing. Description A minimalist fullscreen video player designed for CRT televisions with joystick control. Offers loop mode, volume control, 10-second skip, and a large CRT-optimized start screen. Controlled via joystick buttons/analog stick or
MD2Word – Markdown to Word Converter
Type: Software Status: V.1.0 Tech Stack: Python, markdown-it-py, python-docx, tkinter, PyInstaller, Docker, mmdc/PlantUML/Graphviz (optional) Problem Statement AI assistants and developer tools primarily generate documentation in Markdown. In many organizations — especially in the public sector and enterprises — Word (.docx) is the standard format. Existing converters lose formatting or cannot process embedded diagrams (Mermaid, PlantUML, Graphviz). A simple, locally executable tool that converts Markdown including diagrams into professional Word documents is missing. Description A tool for converting Markdown files into professionally formatted Word documents (.docx). Supports text formatting, headings, lists,
NFCryptoAuth – NFC Authentication & Cryptographic Signing
Type: SoftwareStatus: POC Tech Stack: Python, Flask, SQLite/PostgreSQL, ECDSA (python-ecdsa), Web3/Blockchain (Polygon/Ethereum), ReportLab (PDF), JWT, Gunicorn, Docker, Nginx Problem Statement Art forgery and lack of provenance tracking have been problems in the art market for centuries. Existing certificates of authenticity are easily forgeable, and the chain of custody is usually documented only on paper. Even in the digital age, there is no standardized, tamper-proof bridge between physical artwork and digital verification that operates without a central authority. Description A professional system for authentication and provenance tracking of artworks and valuables

OUTPOST Tactical – Offline AI Field Intelligence
Type: Software Status: POC Tech Stack: Python, React, Ollama, Leaflet/MapLibre, SQLite, FastAPI, PWA, Docker Problem Statement In crisis, disaster, and field deployment scenarios, internet connectivity is frequently unavailable. Existing AI assistants, map services, and knowledge databases are entirely cloud-dependent and therefore useless in offline operation. An integrated, offline-capable platform combining AI support, map material, medical knowledge, survival techniques, and tactical information on a single device — without network dependency — is missing. Description An offline-capable field intelligence platform providing AI-powered decision support even without internet connectivity. Comprises 16 modules: local
OwnBill – Wiederkehrende Services & Rechnungsverwaltung
Type: Software Status: Alpha Tech Stack: Next.js 16, React, Tailwind CSS, Prisma ORM, MySQL, NextAuth, Zod, Docker Problem Statement Freelancers and small businesses need a simple, self-hosted invoicing system without monthly SaaS fees. Existing solutions are either too complex (ERP systems), cloud-dependent, or lack automated management of recurring invoices with configurable templates and email delivery. The combination of customer, service, and invoice management with automatic reminders is missing in lean self-hosted tools. Description A web application for managing customers, services, and invoices with a focus on recurring billing. Offers CRUD

Panacuda Hunters – Livestream of Art, Technology, and Playful Exploration
Panacuda Hunters is our ongoing livestream where we bring together art, technology, and playful exploration. It is an open laboratory, a digital studio, a space where things can take shape – and sometimes fall apart. We don’t only show results; we show the path that leads to them. And that path is rarely linear. In the streams, we discuss current topics or work live on projects, objects, and ideas. We code, model, design, construct, improvise, research, and talk things through. Sometimes what emerges are small, fleeting moments; sometimes they become
Ralphy – Autonomous AI Coding Loop
Type: Software Status: In Development Tech Stack: Bash, Claude Code (CLI), Obsidian (Markdown/REST API), n8n (Docker mit PostgreSQL/Redis), Telegram Bot, Git Problem Statement AI code assistants like Claude Code require continuous human interaction and lose context after each session. For clearly specified development tasks (bug fixes, feature implementations, refactoring), this real-time accompaniment is unnecessary and inefficient. A framework is missing that runs AI development autonomously overnight, documents progress, escalates on blockers, and delivers finished results in the morning — like a night-shift developer. Description A framework for autonomous overnight AI
Screener – Multi-Screen Show Runner
Type: Software Status: Beta Tech Stack: Python, PyQt/Qt, mpv, PyYAML, psutil (optional) Problem Statement Exhibitions, events, and storefronts frequently need multiple screens with rotating, synchronized content. Existing digital signage solutions are either cloud-based, expensive, or lack support for custom Python applications as content sources. Simultaneously controlling different media types (videos, images, webcam, generative art) across multiple monitors with seamless transitions typically requires manual per-screen configuration. Description A flexible system for simultaneously driving multiple screens with diverse content — images, videos, webcam feeds, and custom Python apps (particles, Matrix effect, Game
SOMA – Somatic Operating Model Architecture
Type: Software Status: POC Tech Stack: Python, FastAPI, Uvicorn, psutil, httpx, Pydantic, Ollama, InfluxDB 2.7, React 18, Recharts, Tailwind CSS, Vite, Docker Problem Statement Current AI systems operate completely detached from their physical infrastructure. Even when a server is under extreme load, overheating, or running low on memory, the AI responds identically as when idle. There is no feedback channel between hardware state and AI behavior — no equivalent to fatigue, stress, or exhaustion. This leads to suboptimal resource utilization and prevents emergent self-regulation. Description An experimental Embodied AI framework

SpeakingAI – Interactive AI Avatar with Voice Output
Type: Software Status: POC Tech Stack: Python, OpenAI Assistants API, Pyglet, PyAudio, SpeechRecognition, tkinter Problem Statement AI assistants communicate almost exclusively via text. For exhibitions, events, and interactive installations, a visual presence is missing — a ‘face’ that reacts to questions and shows human-like animations. Existing avatar solutions require elaborate 3D engines, professional motion capture, or cloud rendering. There is no lightweight solution that provides an animated, speaking AI avatar with real-time interaction on standard hardware. Description An animated AI avatar system with real-time voice output and robotic audio effects.
SpotifyMCP – Spotify Control for Claude
Type: Software Status: Beta Tech Stack: Python, FastMCP, httpx, Pydantic, Spotify Web API (OAuth 2.0/PKCE) Problem Statement AI assistants can talk about music but cannot directly interact with music services. There is no standardized interface (MCP) that allows an LLM to control Spotify — search songs, create playlists, or control playback. The integration requires OAuth flows, token management, and API abstractions that are not manageable for end users. Description An MCP server (Model Context Protocol) for full control of Spotify via the Web API directly from Claude. Provides 20 tools:
Stanley – AI Stream Narrator
Type: Software Status: Alpha Tech Stack: Python, OpenCV, OpenAI Vision / Claude Vision, ElevenLabs TTS, Flask, Redis (optional), tkinter, Docker Problem Statement Livestreams suffer from monotonous phases where little happens, and streamers must constantly moderate themselves. There is no AI solution that visually understands stream content and independently commentates — especially none that recognizes people, builds context over time, and speaks in various entertaining personas. Existing stream bots only react to chat commands, not to visual events. Description An AI-powered real-time narrator for livestreams that captures webcam frames, recognizes faces,

The Pannonians (Co-Founder)
The Pannonians are a distinctive artist collective based in the heart of Burgenland, Austria, inspired by a playful narrative of cosmic travelers landing in Pannonia to introduce new artistic perspectives. This concept is embodied in their signature style, the “Pannonian Cut.” The collective’s works span diverse artistic forms, seamlessly integrating modern digital technologies. This innovative approach creates a bridge between physical and digital realms, fostering connections across artistic mediums and eras. Comprised of a diverse group of artists, each member contributes unique skills and perspectives, enriching a collaborative environment that
Trad0r – AI-Powered Crypto Trading
Type: Software Status: Prototype Tech Stack: Python, Flask, SQLAlchemy, SQLite, Binance API, OpenAI/Claude/Ollama, Telegram Bot API, HTML/CSS/JS Problem Statement Crypto markets operate 24/7 with extreme volatility. Manual market observation and trading decisions are time-consuming, emotionally stressful, and error-prone. Existing trading bots are based on rigid rule sets without contextual market analysis. The integration of modern LLMs for qualitative market assessment into the trading process is largely missing. Description A modular trading application with Binance API integration for AI-powered market analysis and automated trading. Offers trailing stop-loss, OCO orders, dollar-cost averaging

Transcript – Audio Transcription App with Speaker Diarization
Type: Software Status: POC Tech Stack: Python, FastAPI, SQLAlchemy (async), Celery 5, Redis 7, PostgreSQL 16, faster-whisper/WhisperX/OpenAI Whisper, PyTorch (CUDA 12.6), React 18, Vite, TypeScript, Docker Problem Statement Transcription of meetings, interviews, and conferences requires not only speech-to-text but also speaker recognition (who said what?). Cloud-based services are expensive and problematic from a data privacy perspective (GDPR). Local alternatives rarely offer comparable quality, no provider comparison, and no admin dashboard for monitoring quality and performance of different Whisper variants. Description A complete transcription platform with three Whisper variants (faster-whisper, OpenAI-Whisper,

TYPO3-Manager
Type: Software Status: In Development Tech Stack: Python, Flask, MySQL, SQLite, Jinja2, MCP (stdio/SSE), REST API Problem Statement TYPO3-based websites, especially in public and institutional sectors, suffer from the same problems as WordPress installations: fragmented administration, no central control across multiple instances, and labor-intensive manual content maintenance. TYPO3’s complex database structure with nested content elements and page tree hierarchies makes direct automation without specialized tooling extremely difficult. Description A Python web application for direct management and automation of TYPO3 websites via MySQL database access. Offers multi-site management, page/post/menu administration, task
WebStore – Point-of-Sale for Physical NFT Art Prints
Type: Software Status: Alpha Tech Stack: Python, Flask, SQLite/PostgreSQL, ThirdWeb SDK, Web3, HTML/CSS/JS (Vanilla), Docker, Nginx, Redis Problem Statement Selling NFT art at events and exhibitions fails due to complexity: customers must set up wallets, buy cryptocurrency, and mint NFTs themselves — a process that completely excludes non-tech-savvy buyers. At the same time, there is no bridge between physical art print and digital NFT ownership. No point-of-sale system unifies both worlds in a single, frictionless purchase process. Description A sales system that mints NFTs while simultaneously creating physical art prints
