4M documents processed
For DVS AG, we built a system that processes varied invoices and receipts, extracts relevant data, and returns it in a structured format. We continue to operate and monitor the solution after go-live.
View case study
We automatedocument-intensiveprocesses with AI.
From initial process analysis to ongoing operation. More than four million documents processed, engineered and hosted in Germany.
- 4M+
- documents processed
- Germany
- engineering, hosting and operation
Previously at
Featured client project
A productive system running in day-to-day operations.
Our core fields
Three building blocks for systems that can be operated reliably.
Agentic AI Systems
Agents for clearly defined domain processes, with human approvals and complete logging.
Document Intelligence
Recognize, classify, validate, and pass documents to existing systems in structured form.
Operations & Compliance
EU hosting, monitoring, audit logs, and contractually defined data processing.
Our products
Software we developed ourselves and continue to operate.
Document workflows without manual follow-up
MandantLink identifies missing documents, contacts clients, and keeps the processing status transparent for the tax firm.
Visit website
Agent2 enables teams to build and evaluate AI agents for clearly defined domain processes.
Visit website
Kontaktscanner finds companies through Google Places, enriches contact data, and exports the results for further processing.
Visit websiteThe team behind Artesiana
We are a three-person product and engineering team. We build the systems ourselves and remain responsible after go-live.
Leo Naderi
Co-Founder | Product, Engineering & Delivery
Leads architecture, technical delivery, and the engineering team. Guides client projects from process analysis through ongoing operation.
LinkedIn
Luca Zacharias
Co-Founder | AI & Quality
Leads models, agent logic, and quality assurance. Translates domain requirements into systems that can be operated reliably.
LinkedIn
Aly ElSharaby
AI Engineer
Develops and tests AI agents and integrations for client projects and our own products.
LinkedInFrequently asked questions
Exclusively in Germany or the EU. On request fully on-premise in your infrastructure – including the AI models.
A first productive pilot is usually ready after 2–4 weeks. From there we iterate in short cycles up to the full rollout.
Directly with the founders, no middle layers. Short paths, weekly demos, and full transparency on progress and cost.
It depends on the process, data, and integrations. After an initial assessment, you receive a transparent proposal covering scope, approach, and cost.
Do you have a process that creates too much manual work?
Tell us briefly how it works today. We will respond personally and tell you openly whether we can help.
