Work
Three anonymized projects — industry, scale, metrics from real operations. Customer names available on request.
Three AI systems, seven stores, six months.
3 production systems · 7 stores
Regional grocery retailer with 7+ locations: store communication ran entirely by phone, staff planning in Excel, facility inspections on paper. Scaling to further stores was impossible without infrastructure.
Voice bot for store communication as a central interface, data-driven staff forecasting based on historical revenue, automated inspection management with a traffic-light system — architected as one coherent system on a shared data layer.
Voice assistance live in all 7 stores within 6 weeks. Staff forecasting automated. 100+ inspection appointments per month in the traffic-light system. The system scales to new stores without re-architecture.
From email to ERP in under 90 seconds.
<90 sec email → ERP entry
Mid-sized packaging and logistics supplier received orders as unstructured email attachments (PDF, Excel, scans). Every order entry consumed staff time — manual-transfer errors were systemic.
AI-driven extraction of order data from email attachments (various formats), normalization into a unified schema, direct ERP integration via PowerShell/EWS. Manual entry is eliminated; humans stay in the approval loop.
Email-to-ERP pipeline fully automated. Customer confirmation email in under 90 seconds. Manual entry reduced to edge cases. Staff focus on optimization instead of data entry.
Automatic detection of order numbers.
73% recognition rate ArUco + vision
Prefab-house manufacturer with nationwide production: metal panels are marked with order numbers that must be captured manually at every station. Errors propagate down the line.
Computer-vision pipeline with camera capture, ArUco marker tracking, and OCR evaluation. Results flow directly into the production database. Planned upgrade path to a local vision model (Qwen3-VL) via Brane for higher recognition rates.
Camera to database entry in under 5 seconds. 73% recognition rate as an honest starting point — manual labeling at the line eliminated for recognized cases. Iteration toward a local vision model as the next milestone.
Industries
Where AI becomes productive with us.
Industry & manufacturing
Email-to-ERP pipelines, vision QA, predictive-maintenance assessments.
Retail & e-commerce
Voice assistance, staff forecasting, product content automation.
Financial services
Document extraction, compliance checks, asset-reporting assistants.
Logistics & mobility
Route planning, document OCR, fleet monitoring.
Public sector
Citizen-information assistants, document classification with GDPR ownership.
Health & life sciences
Radiology pre-processing, document review, terminology normalization.
Your project isn't listed?
We tell you honestly when a use case is production-ready and when it isn't. Free initial call within 48 hours.