Where does AI deliver real value?
In tasks like repetitive visual inspection, heavy document search, first-line answers to customer questions and video-based counting, AI significantly reduces manual effort. First, the process that would benefit most from automation is identified; then results are measured with a small pilot.
For projects that require data privacy, models can run on-device (CoreML) or on the company's own server.
What I offer
LLM assistants & chatbots
Smart assistants for websites and internal systems.
RAG
Accurate, sourced answers from internal documents.
Computer vision
Defect detection and OEE on the production line.
Customer counting
Camera-based store traffic and conversion analytics.
License plate recognition
Parking and traffic monitoring systems.
On-device AI
Private models that run offline with CoreML.
Technologies I use
Official documentation and resources
Standards and official documentation I follow in projects:
Example projects
A selection of live projects serving real users.
Ganicus AI — An AI Employee for Sales Teams
B2B lead finding, map scanning, buying signals, lead scoring, CRM integration and personalized messages in 80+ languages.
View
Inspectra — AI Quality Control & OEE
Edge-AI iOS app for camera-based visual quality inspection.
View
Forbest AVM Pro — Retail Analytics
AI-based visitor counting and footfall analytics.
ViewVELUX License Plate Recognition & Traffic Analytics
Smart plate recognition and traffic monitoring system.
ViewHow we work
Free intro call
We clarify needs, existing systems and goals together.
Quote & roadmap
Scope, technology, timeline and budget within 24 hours.
Staged development
Weekly demos and a live test environment.
Launch & support
Go-live, training and ongoing maintenance.
Free pre-analysisThe first meeting and roadmap are completely free, with no obligation.
Staged delivery and approvalYou review each module on a test environment before we move on.
Written scope and rightsScope, delivery and source code rights are defined in the contract.
Frequently asked questions
How long does AI integration take?
A pilot module is usually up and running in 2–6 weeks, then scaled based on results.
Will my data leave my premises?
Depending on your needs, models can run on-device or on your own server so data doesn't have to leave.
Which AI models do you use?
OpenCV, YOLO, MediaPipe and CoreML for vision; LLM APIs such as OpenAI and Claude for text.
Can AI be added to my existing software?
Yes; it is added to your existing system as an API or microservice.