Expertise 04
AI & Automation Beyond the demo.
North Radius builds AI applications that work in practice: document processing, assistants that give employees the right answer quickly, and automation of recurring processes. We work for businesses in the Netherlands, Belgium and beyond, using language models such as Azure OpenAI, with measurable quality, attention to privacy and a human in the loop where it matters.
What this service covers
AI that really helps.
“AI assistants that know your own documents, smart document processing and automation of repetitive work. Losing sleep over manual work is a thing of the past.”
AI assistants (RAG)
An assistant that answers questions from your own documents, with sources, inside your own software or workplace.
Document processing
Automatically recognising, checking and routing data from contracts, invoices or forms.
Agents & workflows
Automating multi-step tasks, with clear boundaries and checkpoints for people.
Evaluation & monitoring
Agreed quality measures, test sets and monitoring, so you know how well it really works.
My approach
No vague promises or complicated jargon, but transparent communication and solutions that really work.
-
Start with the problem
AI is a means, not a goal. Sometimes a simple rule beats a model.
-
Measure before you build
We agree up front what 'good enough' means and test every version against it.
-
Human in the loop
Where mistakes have consequences, a person stays in control of the outcome.
Technology
What you get.
What you'll have at the end, and the technology I build it with.
- A feasibility check or proof of concept with measurable results
- An AI application in production, connected to your systems
- An evaluation set and quality report
- Monitoring and clear maintenance agreements
- Python Language
- Azure OpenAI AI
- LangChain AI
- Azure Cloud
FAQ
Good questions. Short answers.
What can AI actually do for my organisation?
Common uses are reading and checking documents, an assistant that answers questions from your own knowledge, and automating recurring steps in a process. In an intro call we look at whether AI genuinely adds value in your case.
What about privacy and confidential data?
I prefer models and hosting within the EU, such as Azure OpenAI, and send no more data to a model than necessary. Access, logging and retention are agreed up front.
How do I know the AI solution works well enough?
By measuring it. Together we define a test set and quality measures, such as accuracy and completeness, and assess every version against them, with automated checks and human review where needed.
Can we start small?
Yes, and I usually recommend it. A well-scoped proof of concept shows within a few weeks whether an application is feasible, before you invest in a full solution.
Tell me about your project
Ready for IT that actually works?
A no-obligation intro call takes half an hour. In return you get honest advice and a concrete price indication.