AI Maturity Model for Danish Companies
A practical model that helps you get an overview of where your company is in the AI journey – and what the next safe step is. The model is developed with the EU AI Act in mind.
Why AI maturity is decisive today
Many companies find that employees experiment with AI tools on their own. That brings quick wins, but can also create challenges:
- The value disappears when the employee goes home
- Inconsistent quality and output
- Increased security and compliance risks
- Lack of scalability across the organisation
With the new EU AI Act, the requirements for risk management, transparency and documentation are growing. A structured approach helps you avoid wasted resources and ensure compliance.
This model gives you a clear picture of where you stand, and a safe way forward.
A natural development - not a legal ladder
Our model describes the natural progression a company goes through when employees begin to use AI tools in their daily work.
The progression happens step by step. First, AI is used for individual tasks, then to improve workflows and create more coherent processes. As experience grows, new opportunities arise to automate, standardise and eventually develop solutions together.
Each level builds on the previous one. It is not a series of isolated projects, but ongoing development driven by experience, learning and improvement.
Below you can expand each level and see what it means in practice. The EU AI Act roles are explained separately afterwards, because they always depend on the specific solution.
What characterises your level?
Expand the levels to see their characteristics, examples and next steps. Once you have an overview of how you use AI, you can assess your role under the EU AI Act.
1
Level 1AI tools
Employees use AI actively in daily work - both standalone tools and AI embedded in existing systems.
An AI tool is any piece of software in which a language model or other AI performs work for an employee. This includes both direct AI tools, where the AI is the product itself (ChatGPT, Claude, support agents, email agents), and indirect AI tools, where AI is built into software you already use (SaaS solutions such as ServiceNow, Microsoft 365 and Atlassian). The latter category is often overlooked - but in the eyes of the law, you are a deployer in both cases.
At this level: Employees actively use AI tools to perform their daily tasks - not as an experiment, but as a natural part of their work.
Examples:
- ChatGPT, Claude, Gemini - general text, analysis and sparring
- Image and video generation - Midjourney, DALL·E, Sora
- Copilot in Microsoft 365 - email, documents, spreadsheets
- AI agents in ServiceNow, Atlassian, Salesforce - support, ticketing, CRM
- Claude Code, GitHub Copilot - software development
How we help you move forward:
- Create a shared direction for AI in the business
We help you establish a shared foundation so employees use AI effectively, responsibly and with a focus on tangible business results. - Identify where AI creates the greatest value
We map your opportunities and help you prioritise the areas where AI can improve productivity, quality and workflows. - Ensure secure and scalable access to AI
We advise on platform choices, security and data handling so AI can be used within the right framework from the outset.
2
Level 2Skills economy
Knowledge becomes shareable, improved skills with your terminology, quality requirements and safeguards (guardrails)
On the surface, a skill is a reusable prompt that is continuously improved. But in reality it is much more: Where a language model contains only raw knowledge, a skill represents professional experience - your terminology, your quality requirements, your way of doing things, packaged with guidelines and safeguards. A skill is organised knowledge that can be handed over, shared and improved.
At this level: Employees actively share skills with each other - and more importantly: they actively write and improve their own. Knowledge moves from people's heads into an organised, reusable format.
Examples:
- Proposal writing in the company's tone and structure
- Code review according to internal development standards
- Quality assurance of customer communication before sending
- Report generation with fixed formats and data sources
- Onboarding assistants that know your processes
How we help you move forward:
- Turn the company’s knowledge into a scalable resource
We help you turn employees’ experience and ways of working into reusable AI capabilities that the organisation can build on. - Create a shared model for knowledge sharing and improvement
We establish structures where skills can be developed, quality assured and shared across the organisation. - Embed AI knowledge as an organisational capability
We help you move valuable knowledge from individuals into shared working methods that can be used widely.
3
Level 3Agentic workflows
Employees build and improve workflows where agents perform longer-running or repeated work.
An AI process resembles a traditional work process - except that it uses agents to perform longer-running or heavily repetitive work. Think of it as employing a skill (see level 2) to do a piece of work - and then continuously carrying out performance evaluation and improvement, just as you would with an employee.
User starts process → AI agent retrieves data from the web and performs analysis → review: is the quality good enough? → No: loop back and improve → Yes: result is delivered to the user
At this level: Employees actively build, extend and improve workflows. AI is no longer only a tool you talk to - it is processes that work.
How we help you move forward:
- Turn workflows into intelligent AI processes
We identify opportunities and develop workflows where AI agents can take on complex or repetitive tasks. - Ensure quality and control in AI processes
We build governance, checkpoints and guardrails into solutions so they deliver stable results and can be used safely in practice. - Create AI processes that can scale
We help you develop robust solutions that can be maintained, improved and extended to more areas of the organisation.
4
Level 4Autonomous workflows
Workflows run automatically with monitoring, governance and access controls - without direct supervision, but never without control.
This level is the natural development of level 3: A workflow that an employee has previously started daily moves to automatic execution. This happens when workflows are systematised and integrated with reporting, monitoring, governance and access controls - so the process can run without direct supervision, but never without control.
At this level: The company has an infrastructure for operating and maintaining its own workflows, and the workflows that run automatically have been developed and tested before they were released. People move from performing work to designing and controlling.
How we help you move forward:
- Make AI ready for operations at greater scale
We help you establish the technical and organisational frameworks that make automated processes stable and secure. - Establish governance around automation
We create clear frameworks for ownership, monitoring, maintenance and continuous improvement of automated processes. - Give management control and insight
We ensure transparency through monitoring, reporting and control mechanisms, so automation happens with confidence.
5
Level 5Collaborative engineering
The whole organisation improves AI-integrated components and workflows on a shared platform.
The hallmark of a fully agentic company is that it has built its own platform for governing and improving all AI-integrated components and workflows - one shared place where the whole organisation works.
Think of a kanban board where the company's domain experts set up tasks and explain their intention - and from there build the workflows that the company runs on. An agent takes the task, performs analysis, designs a solution, implements it, performs self-review - starts over if it went wrong, or otherwise closes the task. The domain expert provides knowledge and direction; the agent provides the work.
At this level: All company employees work through such a tool - and the company improves it continuously (continuous improvement). Workflows are self-running, self-maintaining and self-improving.
How we help you move forward:
- Create the foundation for an AI-driven organisation
We help you establish the platforms and structures that make it possible to develop, govern and scale AI solutions across the business. - Build a culture of continuous improvement
We help the organisation work systematically with AI, where employees, domain knowledge and technology continuously improve processes. - Turn AI into a strategic competitive advantage
We help you integrate AI as a permanent organisational capability that strengthens efficiency, innovation and decision-making.
Two roles assessed per solution
The role does not follow the maturity level. You may be a deployer, a provider or both, depending on the specific AI system, purpose and responsibility.
EU AI ActDeployer
When you use an AI system in your organisation and are responsible for how it is used.
The EU AI Act, Regulation (EU) 2024/1689, distinguishes between several roles, including provider and deployer. An organisation using an AI system in its own operations may be a deployer. The specific obligations depend on the AI system's classification, purpose and use.
If the organisation uses a high-risk AI system, specific requirements apply to the deployer role, including:
- Use in accordance with instructions - using the AI system in line with the provider's instructions and the established framework.
- Human oversight - ensuring that people responsible for the AI system have the necessary competence, training and authority to monitor and intervene.
- Relevant input data - where the organisation itself controls input data, ensuring it is relevant and sufficiently representative for the intended purpose.
- Monitoring and documentation - ensuring that the organisation follows the system's use, documents relevant matters and responds to serious incidents or risks.
- AI literacy - ensuring that people working with AI systems have a sufficient level of AI competence.
How to use the model
The five levels show different stages of the organisation's AI maturity and can help identify areas relevant to the deployer role.
The model is a practical tool for creating an overview of AI use, responsibility and governance - not a legal classification.
The specific role and requirements always depend on the individual AI system, its use and the organisation's role.
EU AI ActProvider
When you develop, commission or put an AI system into service under your own name.
When an organisation develops, commissions the development of or places an AI system on the market under its own name or trademark, it may have a provider role under the EU AI Act. The specific obligations depend, among other things, on the system's classification, risk level and use.
For high-risk AI systems, provider obligations can include:
- Risk management - a documented risk-management system throughout the system's lifecycle
- Technical documentation - a description of the system's purpose, design, performance and limitations
- Quality management - processes for development, testing, changes and documentation
- Testing and validation - assessing whether the system meets applicable requirements before it is placed on the market or put into service
- Conformity assessment - where the Regulation requires it for the relevant system category
- Post-market monitoring - following up on the system's performance and serious incidents
How to use the model: Provider circumstances can arise at all five levels. The specific assessment should start with the solution, its intended purpose and the chain of responsibility.