Scotland’s AI Strategy 2026–2031 sets out an ambitious five-year plan for responsible AI in public services. The strategy is right to aim high. The harder question is whether the data and infrastructure behind those services are ready to support it.
Before discussing software, platforms, AI and automation, we need to talk about infrastructure. The underlying architecture is what truly changes costs and experience, not just the interface on top.
We’re due a defrag: stop building the same thing
Nowhere is this more obvious than in licensing, permitting and registration: the processes businesses need to operate legally and safely, and that citizens depend on for decisions that affect their daily lives. Each council and governing body typically buys its own software, works in isolation, defines categories differently and stores data in incompatible formats.
The result is no shared insight, no comparable data, repeated costs and duplicated admin. Delays to inspections, bad actors thriving, and the citizen experience suffers.
An AI model uses whatever data it is given. Feed it inconsistent, siloed or messy data and you will not get a smarter service. You will make mistakes more quickly and with dangerously more confidence.
Digital maturity starts with data
In GovTech, security is necessarily strict, risk tolerance is low, and procurement takes time. Dropping AI into a broken process will not fix it. If a workflow takes six hours because data is unusable, duplicated or approvals are backlogged, AI will not solve that.
Clean the data first, automate second. Only then should you layer intelligence on top of a foundation that can hold it.
We use the term Authentic Intelligence deliberately when we speak about AI here. It is not about deploying AI because it is available. It is the human judgment, domain expertise and local knowledge that have to sit on top of any automated system.
In a licensing workflow, that means enforcement officers who know what a compliant application looks like, councils that understand their local context, the laws and legislative requirements, and infrastructure clean enough that the AI is working with accurate data rather than amplifying errors. Technology makes predictions. People decide what to trust and how to act.
AI can stand for more than Artificial Intelligence. Authentic Intelligence is what makes those systems worth building in the first place.
AUX: designing for agents
Getting the data right is the foundation, and to be frank, a given we all know. Agentic user experience (AUX) is the next goal.
AUX is not about removing humans from the process entirely. It is about removing the admin that stops them doing their job, and allowing that authentic intelligence to be put to good use.
Designing for agents means exactly that: users can access, query and act through a system without ever having to be inside it. No more logging in, manually navigating, pulling data, clicking through screens and re-keying information that already exists somewhere. The system comes to the user, not the other way around.
For an enforcement officer, that means the platform surfaces the right checklist for their next inspection, pre-populated with the business’s history and flagging outstanding actions from other councils before they have even asked. A co-pilot who has already done the prep.
For a business, it means an application process that guides or simply does the submission, catches errors before they cause delays and keeps the applicant informed without them having to chase. For a citizen, faster decisions made on better data, communicated clearly, without navigating bureaucracy they should not have to understand.
In short, you will never have to fill out a form again. None of this is science fiction. But none of it is possible if the data underneath is siloed, unstructured or duplicated. AUX is not a feature you bolt on. It is what emerges when you have built the foundation properly.
What we have learned building it
ePass runs across all 32 Scottish local authorities serving 15,000 businesses to date.
That came from behind-the-scenes work: building shared infrastructure multiple authorities could use, a modular architecture that connects to existing systems, integrations to the wider Scottish and UK digital ecosystem including ScotAccount, Gov Pay and Gov Notify, and consistent licence definitions that previously varied council by council.
The infrastructure is built to serve all residents directly too, not just the businesses that hold licences, and that is the direction in which the platform is heading. A place where everyone living or visiting Scotland can access all their licences in the same place.
When authorities move licensing onto shared infrastructure through ePass, enforcement officers recover significant time previously lost to admin, fraudulent applications fall, and new licence types deploy in weeks rather than months or years. Those results come from fixing the infrastructure layer. The AI capability will follow.
Building on shared infrastructure means the hard work happens once. Every licence type added after that gets faster and cheaper to deploy. That compounding effect is the point. Better infrastructure also changes the experience for residents and businesses. And licensing exists for a reason. The checks behind the taxi you get into, the food you eat, or the event you attend are only as good as the systems running them.
The question worth asking now
The AI Strategy gives Scottish public sector leaders a clear mandate. The question for each authority is how to turn that into action, and the answer starts with an honest look at the infrastructure underneath.
Can you trust your data? Are your systems flexible enough to connect without a full replacement? Do your processes produce structured, consistent outputs that an agent could act on reliably?
If yes, the tools are ready. If not, the best investment is in fixing the basics, because no amount of clever tooling compensates for broken pipes beneath it.
ePass, the shared digital infrastructure platform for licensing, registration and permitting in the public sector, will continue this discussion at the upcoming FutureScot Public Sector AI conference. James Buchan’s masterclass, ‘You’ll Never Have to Fill Out a Form Again: Building the Infrastructure for AI and Automation’, picks up exactly here.
