AI and automation
Kristensson i Skåne AB helps you use AI where it makes a real difference: the tools introduced and governed, manual workflows automated, small internal tools and integrations, guidelines people can follow, and training in practical use.
AI is already present in most organisations. Sometimes through centrally introduced solutions, sometimes through standalone services, and sometimes as features built into systems you already have. So the question is rarely whether to use AI, but where it actually makes a difference and how the use becomes governed, safe and genuinely adopted.
We work in small defined steps with measurable benefit rather than in large programmes. We build in the environment you already have, reuse your existing governance where possible, and start with a use case that matters in daily work. Compliance is a separate area: the AI Act, classification and role assessment belong to the AI Act and AI governance, and that work connects to this one without being the same thing.
At a glance
- Who
- Organisations that want to use AI practically in daily work, and want the introduction to be governed, safe and actually adopted
- What
- Introducing and governing AI tools, automation of manual workflows, small internal tools and integrations, guidelines and working instructions, and training in practical use
- How
- Small defined steps with measurable benefit, built in the environment you already have, reusing existing governance where it fits
Common situations
This is usually how it starts. If any of these sound familiar, we roughly know where to begin.
You have AI licences, but few people use them
We go from licence to adoption: governance, permissions, ways of working and training.
Read about adoption →The same manual work is repeated every week
We map the workflow and automate what is repetitive and rule-driven.
Read about automation →You need a small tool no supplier sells
A defined internal tool or an integration that solves your particular problem.
Read about internal tools →People use AI on their own, without guidelines
Guidelines and working instructions people can follow, tied to your information classification.
Read about guidelines →Offers within AI and automation
Five ways to start. The scope is set by the use case, and they combine.
Introducing and governing AI tools
From licence to actual use: the tool chosen for the task, permissions and sharing reviewed, governance in place, a pilot before broad rollout, and follow-up of what is genuinely used.
You get: a reviewed data foundation, governance settings, a pilot and a way of working for continued rollout.
Scope: set by the size of the environment and the number of usersRead more → AutomationAutomation of manual workflows
We map the repetitive work, decide what is suitable to automate, and build it in the tools you are already licensed for.
You get: a documented workflow, an automation in operation and a description of how it is maintained.
Scope: a defined engagement per workflowRead more → DevelopmentSmall internal tools and integrations
A defined tool or an integration between systems you already have, built for a concrete need rather than procured as a platform.
You get: a working tool, documentation and a handover so that it can be maintained.
Scope: a defined engagement, set by the needRead more → GovernanceAI guidelines and working instructions
Guidelines that say what may be done with which information, tied to your information classification, and working instructions concrete enough to follow.
You get: guidelines, working instructions and a routine for approving new use cases.
Scope: a defined engagementRead more → TrainingTraining in practical use
Training in your own tools and on your own tasks, not a general AI introduction. Participants leave with something they can use the same day.
You get: a training session adapted to your tools and roles, with material participants can return to.
Scope: a defined engagement, per sessionRead more →Frequently asked questions
Is this the same as the AI Act?
No. The AI Act is compliance: which rules apply to your use, what role you have and which use cases need assessing. That sits under the AI Act and AI governance. This area is about using AI in practice. The two connect, but they answer different questions.
Do we need Copilot to start?
No. Automating manual workflows and building small internal tools do not require Copilot, and we are not tied to any single tool. If you already have Copilot licences it is often the fastest place to start, because the benefit sits in work that is already being done.
What happens to our information?
That depends on the service being used, and it is one of the first things we go through. Which information may be used in which tool belongs with your information classification, and the guidelines follow from it.
How do we begin?
With a use case, not a programme. We pick something done often, that takes time and has a clear result, and measure the difference. If it works, we take the next one.
How we work with AI and automation, in detail
The five areas above, described more closely.
Introducing and governing AI tools
Buying licences is not introducing an AI tool. The benefit appears when the tool reaches the work that is actually done, and the risk appears when it reaches information that should not have been shared. So we start with the data foundation: what is overshared, which permissions apply and which information is sensitive. That holds whichever tool it is.
Which tool fits is decided by the task, not by the vendor. If you run Microsoft 365, Copilot is often the closest at hand, because it already sits in the environment where the work happens, and Microsoft’s own guidance puts the data foundation first, before rollout. We also work with other AI services, and we follow the field and try new tools to see which ones genuinely help.
We then set the governance, run a pilot with a defined group and follow up what is genuinely used before going broader. We also develop the ways of working: where the tool helps in your most common tasks, and where it does not.
At a glance
- The tool chosen for the task, not for the vendor
- Sharing and permissions reviewed before rollout
- Governance settings and follow-up of usage
- A pilot before broad rollout
- Ways of working tied to your most common tasks
Automation of manual workflows
Most manual work worth automating has three characteristics: it is done often, it follows rules, and the result can be checked. We start by mapping the workflow as it actually runs, not as it is documented, and decide together with you what is suitable to automate and what should remain a human judgement.
We build in the tools you are already licensed for where possible, because an automation that requires a new platform rarely gets maintained. The workflow is documented and handed over with a description of what happens when something fails.
At a glance
- Mapping the workflow as it actually runs
- A clear line between what is automated and what stays a judgement
- Built in your existing tools where possible
- Documentation and failure handling at handover
Small internal tools and integrations
Sometimes the need exists but the product does not. An internal form that feeds the right system, an integration between two tools that do not talk to each other, a summary someone compiles by hand every month. Such things are too narrow for a procurement and too concrete to wait for.
We build them defined, for a clear need, and hand them over with documentation. The aim is a tool you or we can maintain, not a dependency nobody understands.
At a glance
- Defined tools for a concrete need
- Integrations between systems you already have
- A documented handover
- Maintainable, not an opaque dependency
AI guidelines and working instructions
A guideline that only says be careful governs nothing. What governs is being explicit about which information may be used in which tool, who approves new use cases, and what applies when an answer from an AI tool is used in a decision.
So we tie the guidelines to your information classification rather than writing them standalone, and write working instructions concrete enough to follow day to day. Where you already have approval and purchasing processes we reuse them rather than building a separate AI governance system.
At a glance
- Guidelines tied to your information classification
- Concrete working instructions, not general exhortations
- A routine for approving new use cases
- Existing processes reused where they exist
Training in practical use
General AI introductions rarely change how people work. So we train in your own tools and on your own tasks, with examples from the work the participants actually do.
The training also covers the limits: what the tool is good at, where it gets things wrong and which information does not belong in it. Participants leave with something they can use the same day and with material they can return to.
At a glance
- Your tools and your tasks, not a general introduction
- The limits made as clear as the possibilities
- Material participants can return to
- Tied to the guidelines so theory and practice say the same thing
AI rarely becomes useful through a large programme. It becomes useful when something done often gets easier, and when it is clear what may be done with which information. Would you like to know where that would make a difference for you? Contact us and we will look at a first use case. Read more about how we work and about the AI Act and AI governance.
Would you like to see where AI actually makes a difference for you?
Contact us