Scotland’s AI Strategy 2026–2031 aims to deliver tangible benefits for people, communities, businesses and public services while protecting trust, fairness and transparency.
For many public bodies in Scotland, the immediate challenge is knowing where to start. Across all sectors, 71% of organisations regularly discuss AI at board level, but only 36% have translated those conversations into a formal strategy.
As Practice Lead at Trustmarque Ultima, I see a clear progression: begin by automating routine manual processes, build adoption through tools such as Copilot, and then explore AI agents with appropriate human oversight.
One of the biggest pieces of advice I can give is to stop thinking about the technology. Think about the problems you have or the outcomes you’re trying to achieve. That will lead you in the right direction for where AI can augment what you’re doing today.
It is a simple shift in thinking, but an important one. Instead of looking for somewhere to deploy AI because the technology is available, organisations in Scotland should start with the challenges preventing them from delivering better public services.
Start with the problem, not the technology
Across the NHS, local government and wider public sector, teams are under pressure to do more with the resources and capacity available to them. Before moving towards more advanced AI, organisations can look at the manual, repetitive and time-consuming processes that absorb valuable staff capacity today.
As someone with experience of AI and automation projects, I can reference many examples but one that stands out is a healthcare use case in Wales, where an outdated procurement system meant only one person at a time could work within the system and submit a request. Following modernisation and the introduction of automation, work that had previously taken days could be completed in minutes.
That is the type of practical improvement that can provide the foundation for something bigger, and much more impactful.
The objective is not automation for automation’s sake but through giving people more capacity to focus on work where their expertise and judgement make the greatest difference.
Get the foundations right
Moving from automation into AI also demands solid foundations. Good data, appropriate governance and security need to be considered from the outset rather than added later. But there is another part of the equation that can easily be overlooked: people.
Many organisations already have access to technologies such as Microsoft Copilot but are not necessarily realising their full value, and there are usually two significant barriers to making progress with this and other tools: adoption and confidence.
People don’t always understand how to use it, where it’s going to be beneficial or how to get the most out of it. There can also be a real fear because people don’t know whether they can trust the technology or whether their data is safe.
Technology adoption therefore cannot simply mean deploying licences. Colleagues need the knowledge and confidence to understand where the technology can genuinely help them.
That human element becomes even more important as organisations move towards AI agents.
From Copilots to agents
The next step in the journey is potentially significant. Rather than technology simply helping an individual complete a task, AI agents provide an opportunity to support more complicated processes.
I believe this is where public sector organisations should begin thinking carefully about the relationship between technology and human expertise.
Turning to information requests in healthcare, instead of an individual manually bringing together data from different sources, specialised agents could support different elements of that work, gathering relevant information and preparing it for review.
The human remains firmly in the loop, and the technology could undertake much of the back-office work, while the person reviewing the information uses their expertise to verify its accuracy and ensure the relevant requirements have been met.
The goal of autonomy should not be to remove people from public services. It should be to free people from low-value administrative work so their skills can be focused where judgement, experience and human interaction matter most.
Build confidence step by step
Public sector organisations do not need to leap immediately from manual processes to fully agentic ways of working. A more practical route is incremental: identify a problem, automate an appropriate part of the process, test the impact and expand from there.
For healthcare organisations this means starting with one element of a process and gradually building the agent’s role as confidence and capability increase. It also creates an opportunity to build governance and human oversight into new ways of working from the beginning.
The end goal should always be to turn AI adoption into an operational improvement programme rather than simply another technology deployment.
Innovation will not just come from IT
Perhaps the biggest change AI creates is not technological at all. It changes who can contribute to innovation, and that should mean not just from technical teams but the people who are out there working every single day.
The person who understands where an NHS process causes unnecessary administration, where a council workflow creates delays or where employees repeatedly have to re-enter the same information, may be better placed to identify the opportunity than someone sitting within a central technology function.
The challenge for leaders is therefore to create the environment in which those ideas can surface, while providing the data, governance, security and skills needed to turn them into workable solutions.
Turning ambition into outcomes
Scotland’s AI ambition is bold. Success, however, will be measured in outcomes, not tools. My recommendation, unsurprisingly, is to start with the end goals in mind, not the technology, which I can best sum up as follows:
- Identify the friction and automate the tasks that slow people down.
- Build skills and confidence in tools such as Copilot.
- Put strong data, security and governance foundations in place.
- Introduce agents for more complex work, with people firmly in control.
This is the route from automation to autonomy, with people still at its heart. That is how Scotland can translate AI ambition into better public service outcomes.