Advisory
AI can deliver more. We help you get there.
Some organisations already have what they need, others need one more layer, and some are ready to change how they work. We review where you stand and advise on the next step.
Pick a step
Shadow AI mapped and a policy written, so people know the boundaries they work within. The licences you already have often go further than people think.
Our recommendation Often there is nothing to buy
When needs grow: data processed in your own environment in Europe, costs visible and under control, and contractors working within your boundaries.
Our recommendation Vitinn
A solid foundation, with company data available through questions in plain language. People get more done in a day.
Our recommendation Connect the data
Processes are rethought, not just changed. Sometimes it pays to start over: what would the process look like if it were designed today? Capacity grows without as many new hires, and timely logging prevents revenue leakage.
Our recommendation Rethink the processes
Where AI starts to deliver value
Three levels, and most companies are somewhere between the individual and the team
1·Individual
People use AI for specific tasks, some with connections to systems and a few agents of their own.
The gain stays with the individual
2·Team
A team shares ways of working, for example in software development where AI is used at different stages.
Knowledge builds up within business units
3·Organisation
Processes rethought around AI, shared agents, and AI under the control of the business side.
Value is created across the company
Agents exist at every level. The difference is whether the value they create can be shared.
Where are you on the path?
For companies and public bodies
Four steps. You start where you are.
- 01
Ground rules
Shadow AI mapped and a policy written, so people know the boundaries they work within. The licences you already have often go further than people think.
“Which data may go into AI - and who decides?”
Our recommendation
Often there is nothing to buy
- 02
A secure layer
When needs grow: data processed in your own environment in Europe, costs visible and under control, and contractors working within your boundaries.
“The contractor uses their own tools - but the code never goes through a personal subscription.”
Our recommendation
Vitinn
- 03
Connected data
A solid foundation, with company data available through questions in plain language. People get more done in a day.
“How are we tracking against the budget today?”
Our recommendation
Connect the data
- 04
Value in operation
Processes are rethought, not just changed. Sometimes it pays to start over: what would the process look like if it were designed today? Capacity grows without as many new hires, and timely logging prevents revenue leakage.
“Hours logged as you go are not forgotten - and forgotten hours are never invoiced.”
Our recommendation
Rethink the processes
For software teams
Most teams already use AI. The question is whether it has got further than the editor.
- 01
Assistant
Claude, Codex or Copilot in the editor. Each person gets faster, but the value stays with the individual and ways of working differ from person to person.
“Explain this bug and suggest a fix.”
Our recommendation
Often no further investment is needed - use the licences you already have, such as Codex and Claude.
- 02
Shared ways of working
The tooling used to the full and tied into the team’s processes: specifications, shared skills and quality gates everyone uses. Training brings the team to the same place.
“The team works to the same specification and the same gates - not each to their own method.”
Our recommendation
Training and shared ways of working
- 03
Optimisation
Measure what works and what doesn’t: where changes stall, what they cost and whether technical debt is piling up. Review becomes the new bottleneck and needs to be organised.
“What does each change cost to ship - and where does it wait longest?”
Our recommendation
Analyse and measure
- 04
Agents that do the work
The development process is designed around the agents, not the other way round. They pick up tasks, work in an isolated environment and deliver a proposal with tests. An expert reads the code and decides what ships.
“The bug has a tested fix waiting before the team arrives.”
Our recommendation
Agentic development
How we help
From the first talk to processes in operation. Each service belongs to one step, so you only take what the next step calls for.
The starting point
An assessment of where you stand
We review where you are on the steps: which tools are in use, which data matters and where the value lies. You get a recommendation for the next step, and sometimes it is to buy nothing.
Step 01 · Ground rules
Talks on AI
Everyday use of AI for staff and managers: what it does well, where it fails and how to use it safely.
Step 01 · Ground rules
Advice on secure adoption
Policy and ground rules: which data may be used, where it is processed and who has access. The foundation everything else builds on.
Step 02 · A secure layer
Courses in Claude, Claude Code and GitHub Copilot
Hands-on training built around tasks from your daily work: Claude for staff, and Claude Code and GitHub Copilot for software teams.
Step 03 · Connected data
Getting your data ready
Help getting your data ready for AI: which data matters, where it lives and how to connect it securely.
Step 04 · Value in operation
Processes and agents
We rethink one process from the ground up, from analysis to agents in operation, and measure what it delivers.
Don’t change the process. Rethink it.
Most start by putting AI on top of the processes they already have. The same work gets done, just faster, and the old bottlenecks come along. The biggest value comes when the process is thought through again.
Changing a process
- AI added on top of the existing steps
- The same tasks, just faster
- Steps that existed only because of manual work stay
Example: time logging
People get a reminder to log their hours and AI helps them word the entries.
Rethinking a process
- Start from the outcome: what needs to happen, and why
- Steps that existed only because of manual work disappear
- People work where judgement matters, agents handle the rest
Example: time logging
Hours are drafted from the calendar, work orders and code. The employee confirms at the end of the day, and nothing is forgotten.
Sometimes the best way is to start over. If the process were designed today, with the tools that exist, what would it look like?
We don't start by selling you something
Conversation
We go through where you stand and where it hurts.
Analysis
One process or one team mapped as it works today, and how it would look if it were designed today.
Proposal
What can be done, what can't, what it costs and what it delivers. Sometimes the proposal is to buy nothing.
Delivery
If you go ahead, we take the process all the way. We don't hand over a report.
The outcome comes from the analysis, not before it.
Technology is the smallest part
Successful AI adoption rests mostly on people and processes, and only a little on the technology itself. That’s why we start with where you are, not with the product.
Where do you stand?
A half-hour conversation is often enough to see which step applies and whether the next one costs anything.
Talk to us
Get in touch directly with the people behind this. We answer ourselves.
Andri Örvar Baldvinsson
andriorvar@apro.is
Páll Jónsson
pall@apro.is
Hlöðver Þór Árnason
hlodver@apro.is
Or email us at apro@apro.is