Secure and GDPR-Compliant AI

June 30, 2026

AI
Cloud & Data Platform
Danoffice IT
Security
Case story

Public sector caseworkers often need to review hundreds or even thousands of pages of documentation, legislation, and historical cases before making a decision.
Danoffice IT developed a secure custom AI solution for public administration that combines Retrieval-Augmented Generation (RAG) and a Private Cloud for AI to support decision-making in sensitive cases involving vulnerable citizens and social services.

By combining AI, daily legislative updates, and a fully controlled private AI environment, the solution enables:

✔ In some cases, reduce case preparation from up to a week to a single working day, saving up to 80% of the time spent on research and information gathering.

✔ More consistent and evidence-based decision support.

✔ Reduced risk of AI hallucinations through verified sources.

✔ GDPR-aligned data handling and full control of sensitive data through a Private Cloud for AI.

AI supports caseworkers by processing and surfacing relevant information faster, while legal assessments, professional judgment, and final decisions remain firmly in the hands of qualified professionals.

Can AI Support Sensitive Public Sector Decision-Making?

Cases involving child protection, vulnerable citizens, and social services are among the most complex and sensitive in the public sector.

To make a single decision, caseworkers often need to navigate thousands of pages of documentation spread across multiple systems, historical case files, policies, legislation, and supporting records. Before reaching a conclusion, they must gather relevant information, verify facts, review previous cases, and ensure compliance with current regulations.

This process is highly manual and time-consuming. Preparing a complete case overview can take days, and in some situations up to a week or even a month before caseworkers have the complete picture needed to make a well-informed decision.

At the same time, legislation, directives, and regulations are constantly evolving. Staying up to date with every change while managing a growing workload creates significant pressure on caseworkers and increases the risk of overlooking critical information.

Yet these decisions often cannot wait. Public sector employees are expected to make well-informed decisions quickly, based on accurate information and within a strict regulatory framework.

This was precisely the challenge presented to Danoffice IT when we were contacted by a professional services organization supporting public sector clients. They were looking for an AI solution capable of supporting complex casework while handling highly sensitive personal data in a secure, compliant, and controlled way.

The Challenge Was Clear:

How can AI help caseworkers navigate large volumes of case material more efficiently, enabling them to make informed decisions faster while retaining full control of the decision-making process?

“From a starting point, when using an AI, you upload your material to a public cloud-solution. There will always be a risk related to that. Data can be exposed – you do not have full control. In Danoffice IT, this is a scenario, we always wish to avoid. Therefore, we offer our clients a Private Cloud for AI-solution, a PCAI. Here, all the client’s data are secure and stored on-premises and with full control.”

Leif Elgaard Høj, Director, Innovation & Emerging Technologies at Danoffice IT.

Employees working at computer desks in an office

The Solution: A Secure RAG AI Application for Public Sector Casework

To address the challenge, we developed a secure AI application designed for public sector casework.
The platform enables caseworkers to ask questions in natural language and receive answers based on verified legislation, case records, and approved data sources.

Instead of creating its answers from historical, generic data from anywhere in the World, the AI will now “guess” less and instead answer from facts.

Every response is linked to the original source, allowing users to see exactly which law, regulation, or document the answer is based on. By grounding responses in verified information rather than assumptions, the solution significantly reduces the risk of AI hallucinations.

The platform gives caseworkers faster access to verified information, greater confidence in their decisions, and full control over sensitive data.

The Results: Faster Case Processing, Better Outcomes

Complex public sector cases often require significant time and effort to review. Here are some of the results achieved with the solution:

✔ Case preparation reduced from up to a week to a single working day in some cases, saving up to 80% of the time spent on research and information gathering.

✔ More consistent and evidence-based decision support for qualified professionals.

✔ Reduced risk of AI hallucinations through verified sources and traceable references.

✔ GDPR-aligned data handling and full control of sensitive data through a Private Cloud for AI.

The platform brings together legislation, historical case information, organizational knowledge, and approved data sources in a single AI-assisted workflow. This gives caseworkers access to a broader knowledge base than would be possible through manual review alone, helping them make more consistent and well-informed decisions.
By grounding every response in verified information and linking answers to the original source material, the solution increases transparency, traceability, and confidence in the information provided.

AI Governance, Transparency, and Human Oversight

As AI adoption continues to grow across the public sector, transparency, accountability, and human oversight are becoming increasingly important.

This solution is designed to support caseworkers by helping them find relevant information, legislation, historical cases, and supporting documentation faster. It does not replace professional judgment or automate legal decisions.

Every response is linked to the original source, allowing users to see exactly which law, regulation, policy, or document was used to generate the answer. This helps ensure transparency, traceability, and confidence in the information provided.

AI supports decision-making. Professionals make decisions. Legal assessments, professional judgment, and final decisions remain firmly in the hands of qualified professionals.

The platform is therefore designed to provide clear traceability of AI-generated responses and visibility into the sources used to formulate recommendations.

The purpose of the solution is not to replace public sector professionals, but to help them spend less time searching for information and more time focusing on complex cases and the citizens they support.

Technology Behind the Solution

The application combines several technologies to ensure both accuracy and security.
At the core of the solution is a Retrieval-Augmented Generation (RAG) architecture running within a secure Private Cloud for AI environment. As a municipality or a regional administration, you will not give away your data to a public, generic AI. However, if you have your own AI, in your own cloud-solution – and maybe even placed in your own datacenter, then it is a different matter. The technology behind this type of AI is called A RAG-solution.

The platform uses:

A vector database to store and retrieve relevant content from legislation, case files, and supporting documents
Elasticsearch to provide fast and accurate search capabilities across large volumes of information
A Large Language Model (LLM) to interpret user questions and generate human-readable responses
Secure integrations with approved data sources and legislation repositories
Continuous updates of legal and regulatory content to ensure access to the latest information

When a user submits a question, the platform first searches for relevant information across approved sources. The most relevant content is then provided to the language model, which generates an answer based exclusively on the retrieved material.
This architecture ensures that responses are based on current and verified data rather than relying solely on the model's training data.

RAG AI architecture showing data sources, search, and AI-generated responses.

“The solution is fed with all accessible legislation on these matters – EVERY day. Every single day the database is being scraped, and new data installed. Simply in order to secure, that the answers are 100 % legally correct. With a RAG-solution we can establish a hierarchy for how the solutions must answer the questions from the users. It must go through many layers”.

Leif Elgaard Høj, Director, Innovation & Emerging Technologies at Danoffice IT.

GDPR-Compliant AI for Secure Case Handling

One thing is the solution itself, but what about the users? Can anyone randomly create requests in this system? In this system, clear rules define who can access what, ensuring that all case handling follows strict security, privacy, and compliance standards.

“This is super private information, so this is critically important. First and foremost, there is an arm’s length principle. A given case must only be accessed by the user, who started the case. The cases cannot be shared – and will be deleted after 30 days. Obviously, all social security numbers are erased automatically”, Leif says and adds. “That way, a municipality can now subscribe to different agents, which can handle GDPR, EU Directives and the EU AI Act totally compliant without the employees having to deal with it”.