AI automation
that truly works —
tested, documented,
ready to deploy. Real solutions
for real problems —
measurable results,
from practice. From idea
to finished system —
live AI solutions,
directly transferable. AI in action
no theory —
documented, tested,
ready to use. Practical examples
that convince —
eight case studies,
openly shared. Automation
that convinces —
real results,
no promises.

Invoices were created manually in Excel — time-consuming and error-prone. XRechnung compliance for public clients was missing and had to be laboriously corrected.

• Docker-based system with n8n workflows, Python API and Ollama LLM
• Three input methods: voice input, camera scan, text entry
• Fuzzy matching for customer and product recognition from natural language
• Excel stays as backend — familiar interface, no retraining needed
• Automatic XRechnung generation (EN 16931) for public-sector orders
• Web interface for master data management (customers, products)

Invoice creation reduced from 15 minutes to under 2 minutes. Full compliance with German e-invoicing standards (XRechnung) without additional software.

Handwritten notes and delivery slips from the construction site had to be typed manually. Information was lost or arrived late at the office.

• Camera integration in the web interface for instant scanning
• Google Cloud Vision API for handwriting recognition (DOCUMENT_TEXT_DETECTION)
• n8n workflow for processing: Image → OCR → Structured Data
• Automatic extraction of customer, quantity, product from scan
• Direct forwarding to invoice or documentation

Photograph a delivery slip — invoice created. No media breaks, no lost papers, no more manual typing.

Company knowledge was scattered across documents, emails and people's heads. New employees needed weeks to get up to speed, experts were constantly interrupted with routine questions.

• RAG system (Retrieval Augmented Generation) with ChromaDB as vector database
• Automatic indexing of documents from network drives
• Admin panel for system control: container management, indexing, cache, logs
• Docker-based architecture for simple deployment
• nginx reverse proxy with SSL for secure access
• Live monitoring and health checks

Employees ask the system instead of colleagues. Onboarding time drastically reduced, expert knowledge preserved and always accessible.

In hybrid working life, office occupancy was unpredictable. Employees arrived to find their preferred spot taken — or offices stood empty. Informal WhatsApp groups and email arrangements led to unfair desk allocation. A fair, transparent management solution without extra admin overhead was completely missing.

• Complete web booking system for desks and meeting rooms
• Interactive room plan with real-time availability at a glance
• Quick booking: date, time, spot — confirmed in under 30 seconds
• Employee search: Who is in the office when? Perfect for spontaneous meetings
• Admin panel: room management, building maintenance, equipment features, blocked times
• User role management (Admin/User) with separate admin cockpit
• CSV import for rapid bulk creation of spots and users
• Mobile-optimized interface for bookings on the go

Transparent space utilization, fair desk allocation, end of WhatsApp groups. Facility management can control office space specifically. Employees book in seconds, the office always knows who is coming.

In day-to-day operations, customer data, quotes, invoices, construction site documentation and tender research were spread across isolated tools: Excel lists, Word templates, photo folders on smartphones and manual searches on tender portals. Data was maintained twice, quotes were copied together from old Word files, site photos sat unsorted on phones and matching tenders were simply missed. A unified system connecting all processes was missing.

• Stand-alone ERP/CRM system based on ASP.NET Core (C#) with a modular controller and view structure
• Customer management, product catalog, quote and invoice creation in one continuous process chain
• Construction site management including photo documentation directly within the system
• AI-powered tender module: relevant tenders are loaded, analyzed and ranked by fit automatically
• Integrated RAG module that makes company knowledge, past quotes and service descriptions usable for answers and new quotes
• Role and rights management with dedicated admin area, user accounts and secure authentication
• Deployment on IIS under Windows; clean publish folder with app-offline mechanism for zero-downtime deployments

Six formerly isolated tools (Excel, Word, photo folders, tender portals, mail attachments, notes) were merged into a single system. Quotes and invoices are now produced at the click of a button from the customer database, site photos are automatically assigned to the right project, and matching tenders appear in the dashboard automatically — instead of being missed. Data maintenance is unambiguous, sales throughput is noticeably higher, and the entire company knowledge sits in one central place for the first time.