Research

AI Research

Automate tasks, not jobs

The AI opportunity for Scotland’s public services. Independent analysis identifying the highest value public sector workflows for AI driven productivity gains.

Download full report (PDF 2.9mb)

Executive summary

Scotland’s public services face a defining moment. Scotland’s public services face a defining moment. The 2026/27 Budget and multiyear Spending Review highlight the central challenge of the decade ahead: sustaining service quality in the face of demographic pressure and constrained resources. To meet this, the Scottish Government has mandated a significant focus on efficiency and reform across the entire public sector.

This is not simply a funding problem; it is also a capacity problem. Without releasing frontline time from administrative and coordination overhead, services will struggle to sustain access, quality and equity, particularly in health and social care, education and local government.

This paper sets out the case for adopting AI-enabled service redesign in Scottish public services to reduce administrative burden, improve responsiveness and protect frontline capacity guided by a clear principle throughout - automate tasks, not jobs. The objective is workforce enablement and better outcomes resulting in higher throughput, shorter backlogs, faster response times and more staff time for direct public value work.

What we analysed

We assessed 50 high-volume public sector services across Scotland with the strongest potential to release staff time through credible AI use cases, spanning:

  • NHS Scotland
  • Education
  • Scotland’s local authorities
  • Police Scotland and justice-adjacent administration
  • Other public bodies, including high-volume transactional services

Across these 50 services, we estimate a baseline of ~178 million staff hours per year, distributed approximately as follows:

This chart shows Storm ID’s estimate of the total annual staff time across 50 high-volume public sector services in Scotland. The baseline is approximately 178 million staff hours per year, representing the total pool of time with potential to be released through credible use of AI.

This chart shows the estimated proportion of staff hours attributable to NHS Scotland within the total 178 million annual staff hours across 50 public sector services. NHS Scotland accounts for approximately 80 million staff hours per year, representing the largest single share of the total baseline.

This chart shows the estimated proportion of staff hours attributable to Education services within the total 178 million annual staff hours across 50 public sector services. Education services account for approximately 37 million staff hours per year, representing the second largest single share of the total baseline.

This chart shows the estimated proportion of staff hours attributable to NHS Scotland within the total 178 million annual staff hours across 50 public sector services. Scotland's local authorities accounts for approximately 32 million staff hours per year, representing the third largest single share of the total baseline.

This chart shows the estimated proportion of staff hours attributable to NHS Scotland within the total 178 million annual staff hours across 50 public sector services. Police Scotland and justice-adjacent administration services account for approximately 21 million staff hours per year, representing the fourth largest single share of the total baseline.

This chart shows the estimated proportion of staff hours attributable to NHS Scotland within the total 178 million annual staff hours across 50 public sector services. Other public bodies account for approximately 8 million staff hours per year, representing the remaining share of the total baseline.

Our modelling focuses on capacity release (time back) rather than assuming immediate cash savings. In practice, AI benefits in public services will typically show up first as improved throughput, reduced backlogs, and increased time for higher value work. Budget impacts depend on later workforce and service design choices.

Up to 62.1 million staff hours per year released by 2030
Low, moderate and optimistic adoption and impact scenarios

This chart presents Storm ID’s modelled estimate of annual staff capacity release across 50 high-volume public sector services in Scotland under different AI adoption scenarios. The first scenario assumes low adoption of AI-enabled use cases and average efficiency improvements of up to approximately 16.6 million staff hours per year could be released. The second scenario assumes a moderate adoption of AI-enabled use cases and average efficiency improvements of up to approximately 36 million staff hours per year could be released. The third scenario assumes an optimistic adoption of AI-enabled use cases and average efficiency improvements of up approximately 62.1 million staff hours per year could be released.

Capacity release to absorb demand pressures and improve service
% of staff baseline hours per year

This chart presents Storm ID’s modelled estimate of annual staff capacity release across 50 high-volume public sector services in Scotland under different AI adoption scenarios. The first scenario assumes a low adoption of AI-enabled use cases, under these assumptions approximately 9% staff hours per year could be released. The second scenario assumes a moderate adoption of AI-enabled use cases, under these assumptions approximately 20% staff hours per year could be released. The third scenario assumes an optimistic adoption of AI-enabled use cases, under these assumptions approximately 35% staff hours per year could be released.

Annual time back for public sector workers by 2030
Potential sector-level implications

This chart presents Storm ID’s modelled estimate of annual staff capacity release across 5 public sector categories in Scotland under a high AI adoption scenario. Under these assumptions, NHS Scotland could save up to approximately 27.1 million staff hours per year. Education could save up to approximately 13.9 million staff hours per year. Local Government could save up to approximately 10.8 million staff hours per year. Policing and Justice could save up to approximately 7 million staff hours per year. Other public bodies could save up to approximately 3.4 million staff hours per year.

50 high-volume services prioritised for AI-enabled service reform
The route to scale is to build reusable and configurable AI components

5 repeatable service patterns
Operational workflows cluster into a small number of repeatable patterns

The 5 patterns are: Case and record lifecycle management Application processing and eligibility or decisioning First contact, triage and routing Scheduling, capacity and follow-up orchestration and Knowledge-intensive documentation

Data sovereignty is a requirement
Half of the top 50 services would likely require a private deployment

48% of the 50 services assessed would likely require a private or sovereign AI deployment model, primarily due to high data sensitivity. 46% of services could be hosted on UK hyperscale cloud AI infrastructure. 6% of services may require a hybrid approach depending on specific use cases and data sensitivity.

The paper also shows that demand and administrative load are set to rise over the next few years without reform. AI enabled redesign can do more than marginally improve today’s position: it can help offset projected workload growth if implemented at scale with the right operating model and controls.

The opportunity: repeatable service patterns

While Scottish public services are diverse in mission, their operational workflows cluster into a small number of repeatable patterns. Across the fifty services analysed, these can be grouped as:

Service pattern distribution across the 50 workflows

Document management represents 51% of total hours

This chart shows how baseline annual staff hours are distributed across five repeatable service patterns, based on analysis of 50 high-volume public sector services in Scotland. Case and record lifecycle management accounts for the largest share at 90.9 million hours, representing 51 percent of total baseline hours. Application processing and eligibility or decisioning accounts for 42 million hours, or 23.6 percent. Knowledge-intensive documentation accounts for 31.4 million hours, or 17.6 percent. First contact, triage and routing accounts for 10 million hours, or 5.6 percent. Scheduling, capacity and follow-up orchestration accounts for 3.3 million hours, or 1.9 percent.

Scotland does not need to build fifty bespoke solutions. The fastest, safest route to scale is to build reusable and configurable AI components aligned to these common patterns and integrate them into existing systems of record.

Automate tasks, not jobs

The paper is explicit that the aim is workforce enablement, not indiscriminate headcount reduction. In a system already under workforce strain, the intended outcomes are:

  • More time for direct care, teaching and professional judgement
  • Shorter waiting times and faster decisions by increasing throughput
  • Improved consistency and quality in documentation, correspondence and case preparation
  • Better staff experience by reducing duplicative admin work

Delivering this credibly requires making the approach operational through workforce engagement in the design of new AI services, training on how to adopt new workflows and ongoing monitoring and continual improvement. This will ensure that time saved translates into better outcomes and is not simply absorbed again by unmanaged demand.

Infrastructure and data sovereignty

The paper compares infrastructure models for hosting AI services, including UK based cloud services appropriate for many transactional and administrative workloads, and private / sovereign environments suited to higher sensitivity contexts such as clinical and criminal justice data, where additional privacy, security and access controls are required. The analysis indicates that around half of the top 50 services assessed would likely require a private deployment model based on data sensitivity and risk.

AI infrastructure requirements by service type

This diagram shows the proportion of services that would require a private or sovereign AI deployment model. Approximately 48 percent of the 50 services assessed fall into this category, primarily due to high data sensitivity. This includes services involving clinical data, criminal justice information, and work with vulnerable people.

This diagram shows the proportion of services suitable for deployment on UK-based hyperscale cloud infrastructure. Around 46 percent of the services assessed fall into this category, typically covering transactional services, education-related workloads, and administrative functions.

This diagram shows the proportion of services requiring a hybrid AI deployment model. Approximately 6 percent of services have mixed or flexible requirements that combine elements of private and cloud-based infrastructure.

Call to Action

To turn opportunity into outcomes by 2030, this paper argues Scotland should:

  1. 1
    Prioritise high-volume, lower-risk workflows first.
    Demonstrate value quickly and build institutional capability.
  2. 2
    Build shared, reusable AI components.
    Mapped to common service patterns, avoiding duplication across NHS boards, councils and public-sector bodies
  3. 3
    Establish robust governance and assurance.
    A prerequisite for scale: clear accountability, audit trails, cyber controls, testing and monitoring and defined human oversight which is consistent with Scotland’s commitment to trustworthy, ethical and inclusive AI.
  4. 4
    Adopt a mixed infrastructure strategy.
    Combining UK public cloud for suitable workloads with private AI infrastructure for high sensitivity services.
  5. 5
    Treat workforce enablement as a core deliverable.
    With training, role redesign, and staff engagement throughout.

Scotland cannot bridge its structural capacity gap through incremental digitisation alone. This analysis demonstrates that if implemented safely, transparently and in partnership with the workforce, AI has the credible potential to release tens of millions of staff hours from current administrative burdens by 2030 and help contribute to meeting the £1.5bn of efficiency savings targeted over the Spending Review period.

Download the AI White paper

Learn about the AI opportunity for Scotland’s public services

Download full report (PDF 2.9mb)