Make AI work and lift
your business \
We start from your challenges, goals and strategy. Together we find out where AI can make a difference, get it working in practice and give your own people the experience to carry on.\
\ Forward deployed: our people work together with yours.
\ Results for our clients
- 10 weeksfrom the first prototype to the launch of four news sites and 28 newspapers at Sjællandske Medier. We estimate that it would previously have taken up to a year.
- 1 hourfrom the Product Owner at DSV Transport has a wish for a feature until it can be in production. AI agents in the code base mean that, in principle, anyone can contribute, although a person still approves before production.
- DKK 0in licence fees for an external CMS at Sjællandske Medier. Marketing edits texts and paywalls themselves and sees the change on the real site.
- 1 monthof active work was what it took to deliver Lungeforeningen's member portal. It has reduced the number of manual enquiries to the association.
What we help you do
We help you use AI in management, in operations and in development. You can start where the need is greatest.
In management
Choose where to use AI
Together we turn your strategy and business goals into a few concrete initiatives that you can prioritise, budget for and follow up on.
In operations
Use AI in the daily work
AI as part of the work with cases, orders, customers and data. With clear rules for what happens automatically and what a person approves.
In development
Strengthen your own development teams
Agents working in your code base, automated tests and a way of working where your developers describe the tasks, review the code and approve the changes.
Three steps.
The goal is that you can do it yourselves.
Forward deployed
We work forward deployed. Our people work in your systems and with your data, close to the daily work and often together with your own IT team. We upskill the team and establish the layer that orchestrates the AI agents. That lifts the team and increases the use of AI in every development process.
01
Step 1 · Clarify
Find out where AI can make a difference
We turn your goals into concrete opportunities and give you one clear recommendation, with a basis for the decision that you can stand behind.
Five questions we help you answer
- Where can AI support our strategy?
- What do we do first, and what do we leave?
- Are our systems and data ready?
- How do our developers get more out of AI?
- How do we manage access, responsibility and risk?
Example: Det Danske Spejderkorps. The first step was a small experiment.
02
Step 2 · Try
Try it on a real task
Try the idea before you invest further. We test it on your own data together with the people who will use the solution.
What you take away from a trial
- Prototype on your own data
- Your employees involved from the start
- Tests with real users
- Measurement before and after
- Assessment of operations and security
- Proposal for a decision: go on, adjust or stop
Example: Lungeforeningen. The prototype became the foundation.
03
Step 3 · Anchor
Get the solution into operation, with your people all the way
We get the solution working in the daily business and help your own people run it from there.
What we get working together with you
- AI in workflows
- AI with access to your systems
- AI in your development
- When something needs to be built
Example: DSV Transport. First rules and tests. Then the agents.
Security and responsibility
Guardrails that make it safe to use AI
It must be possible to use AI without putting data, customers or trust at risk. So we build the guardrails in from the start, as part of the solution and not as a document in a drawer.
You decide what AI may be used for
You decide which models and tools are approved, and who may use them for what. Every use is assessed for risk before it gets access.
Confidential data stays where it belongs
Personal data and confidential information can trigger a warning, be stopped, require an approval or be sent to a model that runs in the EU or on your own infrastructure.
You know where AI is used
One register of uses, systems, models and data. Including the tools employees have started using themselves, so they can be assessed and either approved or replaced.
You can explain what happened
It is logged what AI has had access to and done, who approved it, and when a rule has been overridden. People approve what costs money or cannot be undone.
What we take into account
We are not lawyers. You make the legal assessment together with your own advisers. We make sure the technology can live up to it.
For those responsible for the technology
How we build and run the solutions
The platform for Sjællandske Medier is developed and run with AI agents that we orchestrate. Our developers decide, test and say yes or no before anything goes live. That is the way of working we help your team adopt.
Read the technical walk-through in the Sjællandske Medier case- 0lines of code written by hand on the platform for Sjællandske Medier. The agents wrote the code. Our developers were responsible for architecture, testing and approval.
- 221alarms examined by an agent in the first week after launch, before a person was woken. Real errors call the developer on duty with the cause and a fix ready.
- 83%of the code changes in our projects were made by agents in October 2026. We review every change.
- 0releases made by an agent at DSV Transport. Agents build and test. A person releases.
Models and tools we work with
We are not tied to one supplier. We choose to suit the task, your requirements for data and the agreements you already have. Open models such as Qwen can run in the EU or on your own infrastructure.
Claude Code
Codex
Copilot Cowork
Qwen
The people you will talk to
You talk to the same people who are responsible for our projects. They know what happens when a strategy meets a code base, a budget and a busy working day.

Mads Hofman
CEO & Advisor

Andreas Bødker
Founder & Advisor

Louise Dreyer
Creative Director

Hari Carreras Pérez
CTO

Tejs Stubkjær Rimmer
Director of Product Management
Questions and answers
What we are most often asked before a collaboration begins.
Are you a consultancy focused on AI?
We are a digital agency that both advises and builds solutions. We have done so since 2008. AI has become part of our own work, and we use that experience when we advise you and develop solutions together with you.
What does forward deployed mean?
That our people work in your systems and with your data, instead of delivering a solution from a distance. Often we work together with your own IT team: we upskill the team and establish the layer that orchestrates the AI agents. That lifts the team and increases the use of AI in every development process. The term comes from the software industry. For you it means that decisions are made close to the daily work, and that the experience stays with your own people.
Do you still build apps and websites?
Yes. We build apps and websites together with our clients and start from what you want to achieve. From there we find the solution that fits the task.
Can we get advice without having you build anything?
Yes. You can use our recommendation with your own team or with another supplier. After each step you decide whether we should continue together.
We already have a development team or a supplier. Does it make sense to talk to you?
Yes. We can work together with your own developers, in your code base and with your tools. The goal is that they can carry on themselves when we are done.
What if the advice is that we should not build anything?
Then we say so. Our recommendation may be to buy a ready made solution, to wait until the groundwork is in place, or to leave the idea. It has to make sense for you, and we take pride in transparency and well founded recommendations.
What does it cost?
It depends on the scope of the task. Before we begin, we agree the task and a budget frame, so you know what the work covers and what to expect.
Which models do you use, and will we be tied to one supplier?
We choose model and tool to suit the task. Today we work with Claude Code, Codex, Copilot Cowork and Qwen, among others. We build the access to your systems as an independent layer. That makes it possible to change model later, for example to one that runs in the EU or on your own infrastructure.
What about our data and security?
We agree from the start which data AI may see, what is logged and who approves changes. Sensitive data can be stopped or sent to a model that runs in the EU or on your own infrastructure. If we process personal data for you, we enter into a data processing agreement.
Tell us what you want to achieve.
Then we talk about where AI can make a difference for you, and where it makes sense to begin.
"We want to understand the problem before we recommend a solution. So we ask questions, challenge assumptions and give you advice we can stand behind. Also when the advice is not to build."











