02 · AI automation · Logistics
Automated document processing
A logistics provider processed hundreds of delivery notes, waybills and customs documents every day — manually, error-prone, slow.
- n8n
- Claude API
- OCR
- PostgreSQL
- Ollama
- 85%
- less manual data entry
- 4 min → 12 s
- processing time per document
- 99.2%
- extraction accuracy
01 · Problem
Up to 400 incoming documents a day in various formats — PDFs, scanned images, email attachments. Three employees each spent 3–4 hours a day manually entering data into the ERP system. Data entry error rate: approx. 3%, which regularly led to delivery delays.
02 · Solution
vona built a fully automated processing pipeline with n8n: incoming documents are converted to text via OCR, then classified and extracted into structured data by a locally hosted LLM (Ollama + Llama 3). When the model is uncertain, a human is notified to review. The extracted data flows directly into the existing ERP system. Because the model runs locally, no documents leave the company — GDPR-compliant.
03 · Result
85% of documents are processed fully automatically. Processing time per document cut from an average of 4 minutes to 12 seconds. Extraction accuracy: 99.2%. The three employees were relieved of data entry and moved into quality assurance roles.
Related services
We can do this for you, too.
The services behind this project — with examples, process and frequently asked questions.
Keep reading
Background from practice.
Blog
Automation with AI Workflows: A Guide with Checklist
Which processes suit AI automation, how a workflow is built, which tools exist and how to calculate expectations and benefit realistically.
Read more →Blog
AI Agents in Everyday Work: Uses, Limits and Safety
What sets AI agents apart from chatbots and workflows, when which solution fits, which fields of use have proven themselves and which safety rules you need from the start.
Read more →Next project
Your project here?
Tell us about your problem. The first conversation is of course free of charge and without obligation.
Request a project →