References
Six anonymized cases from our service areas. On customer request, industry and concrete numbers are abstracted; the nature of the task remains. Concrete references with logo and quote are available after an initial conversation.
Ticket triage for a regulated mid-sized company
Problem: Service desk with 600 tickets/week; manual sorting consumed two full-time days per agent.
Solution: AI agent classifies incoming tickets into 23 categories, routes automatically, drafts first response.
Outcome: 72% auto-routing rate, mean first reaction from 4h to 35min, NIS-2-compliant audit trail.
Invoice automation in a family business
Problem: ~1,200 incoming invoices/month sorted, scanned, entered into DATEV by hand.
Solution: Document classification + OCR + structured extraction + pre-entry into ERPNext with human-in-the-loop approval.
Outcome: 85% auto pre-entry, GoBD-compliant audit trail, three days/month freed up in accounting.
Content workflow with brand voice control
Problem: Three-person marketing team had to serve blog, newsletter, and social in parallel; brand voice drifted across writers.
Solution: Brand voice as prompt layer, RAG over studies and white papers, n8n workflow from briefing to draft with voice check.
Outcome: Time-to-publish from 9 to 3 days, consistent brand voice, editorial team works strategically again.
Knowledge assistant for service desk
Problem: Onboarding of new service staff took six months; senior staff were constantly interrupted.
Solution: RAG over Confluence, ticket histories, runbooks, SharePoint; ACL-respecting knowledge assistant with source enforcement.
Outcome: Onboarding halved to three months, 40% fewer internal queries, documented knowledge gaps closed deliberately.
CVE prioritization for a KRITIS operator
Problem: 5,000+ open CVEs in a grown infrastructure; patch backlog unmanageable.
Solution: AI workflow combining CVE feed, EPSS score, asset inventory, threat intelligence into weekly prioritized patch recommendations.
Outcome: MTTR for critical patches from 45 to 8 days; NIS-2 audit without findings; clear escalation paths.
End-to-end pipeline for an energy supplier
Problem: Four core systems (ERP, CRM, billing, portal) without clean interfaces; data synced manually each day.
Solution: Event bus with Kafka, FastAPI microservices as adapters, OpenTelemetry tracing across hops, audit trail per transaction.
Outcome: Data consistency from 89% to 99.7%, manual sync eliminated, self-service portal uses live data.
Discuss your project
Seven questions on industry, size, and biggest routines — we respond within a business day with a first assessment.
> Start AI Readiness Check