Scottish Agriculture’s Digital Shift: Cloud Skills, Data Sharing and Satellite Monitoring

The Scottish Government’s Agricultural Economy Directorate is building digital services around a practical constraint: technology has to fit the work it supports. Its platforms help farmers submit claims for subsidies, with the directorate handling about $1 billion in payments. The infrastructure behind those services spans VMware systems, Kubernetes deployments and Amazon Web Services, supported by specialists in systems, databases, identity, automation and cybersecurity.

Neill Smith, the directorate’s Head of Infrastructure, says that mix is a reason to resist treating cloud adoption as a destination in itself. Workloads should run where their requirements point, he argues. That approach leaves the team managing a varied technical estate while it works to make services easier for farmers and adapt its application delivery.

The interview ranges from cloud strategy and recruitment to data sharing, private AI and satellite imagery. The through line is operational: modernization depends on the people who can run complex systems, and the value of a technology comes down to whether it improves a service or removes work from its users.

Cloud strategy starts with the workload

Smith pushes back on a common interpretation of a cloud-first strategy: that every system should move to the cloud. “That’s not what it means,” he says. In his view, the hosting decision should follow the needs of each workload. A large share of the directorate’s systems may end up in the cloud, but that does not make cloud the right answer in every case.

His point also applies to organizations that commit to one provider and then try to reshape their requirements around that choice. Smith questions why a team picked a particular platform if it now has to “shoehorn” its work into the available services. AWS and Azure each offer a broad set of capabilities, he says, but neither provides the best option for every need.

“I would rather begin the conversation by asking, what does my requirement dictate?”

That workload-led approach can mean using more than one environment, including on-premises infrastructure, virtual machines, containers and public cloud. The tradeoff is added operational complexity. Smith says multi-cloud can require skills across multiple providers, at a time when cloud expertise is already difficult to recruit and retain. A technology strategy therefore has to account for the capacity of the team that will manage it.

For Smith, the human side of cloud migration determines whether the change works. “People and culture drive the change,” he says, warning that a poorly handled move can leave an organization in a difficult position. Cloud platforms give teams options, but those options bring new responsibilities for architecture, operations and ongoing learning.

Digital services set a higher bar

People now expect many routine transactions to be quick and straightforward, Smith says, whether they are buying a car or applying for a credit card. Digital services in the public sector face the same expectations. A well-designed online process can make it easier for users while reducing the time and cost involved in handling applications.

For farmers, replacing a paper form with an online one is a useful improvement. Smith sees a larger opportunity in connecting services that still operate separately. When users are asked to enter information an organization already holds, the process creates avoidable effort. Better data sharing could reduce repeated requests and make services feel more joined up.

He points to Estonia as an example of a country that has made extensive use of online services, describing its digital society as an ambitious model. The practical lesson for the directorate is that digitization can go beyond putting an existing form on a screen. It can also change what information a user needs to provide and how different services pass information between them.

Cloud skills are difficult to build and keep

Smith describes cloud engineering and Kubernetes as complex fields that change quickly. Their difficulty helps explain the demand for experienced engineers and the high salaries they can command. He says organizations did not always appreciate how much complexity sat inside the applications they planned to move.

“We’ll just put it in the cloud, and everything works.” And it’s not like that simple.

Traditional infrastructure teams could divide responsibilities among network, backup and storage specialists. Cloud roles can expect an engineer to understand several of those areas at once, while platforms and practices continue to change. That breadth makes it hard for teams to keep their skills current, particularly when they are also responsible for services that must keep running.

Recruitment and retention are central challenges for the directorate. Smith says it has a substantial training budget, which can help bring in people who have interest and aptitude but lack extensive experience or formal certification. With a supportive culture and a suitable environment, he believes those staff can develop quickly.

That strategy carries a realistic risk: some trained staff will leave for technology firms, including VMware and Xtravirt. Smith accepts that attrition as part of the equation. The people who stay can build experience inside the organization, helping it improve its practices and take on new work. Training is valuable, in this account, but it cannot by itself eliminate competition for experienced specialists.

Broader perspectives matter for AI

Smith connects the makeup of technical teams to the risk of bias in AI systems. He observes that technology events where people discuss AI and bias can still have predominantly male audiences. If the teams building large language models also draw from a narrow range of backgrounds, their assumptions may be reflected in the resulting systems, even without anyone intending that outcome.

“If you have more of an eclectic mix of people and a diverse team, the better you’ll have in terms of ideas.”

He says technology remains male-dominated, though he sees gradual improvement. His argument is both about fairness and the quality of problem-solving: people with different experiences can notice different gaps and bring a wider set of ideas to technical decisions. That matters especially when AI systems affect people whose perspectives may not be represented among their builders.

Private AI calls for measured exploration

Smith is interested in private AI, but he describes it as an early area for the directorate. He sees a potential fit for organizations that want to explore AI while keeping close control over how they use it. At the same time, he says the technology is not mature enough for the directorate to rush into broad adoption.

His preferred course is internal exploration: let staff learn what the tools can do, while keeping their use bounded. Smith describes himself as willing to push at the edges of what the organization can do, provided the work stays within its rules. He also points to uncertainty around the use of creative material in AI development and the possibility of reputational harm if an organization handles the subject poorly.

That caution leaves open practical questions the interview does not resolve: which workloads would benefit from private AI, what controls would be needed, and how teams could assess the quality of its outputs. Smith’s position is to keep the topic under examination while waiting for the technology and its operating practices to mature.

Satellite imagery could simplify farm claims

Sustainability is a major focus for the directorate, and Smith says technology could help support greener farming. He describes schemes intended to encourage practices such as protecting habitats and accounting for crop size. Agriculture, he says, is a major source of carbon emissions, making better information about land use relevant to environmental efforts.

The directorate is working with Planet Labs, which captures daily images from the air. Smith says the satellite imagery can resolve features down to three or four centimeters. Combined with AI, the data may help detect changes to field boundaries and crop types, as well as assess soil and carbon emissions.

Today, people inspect fields to check whether reported land use matches what is present. Smith says the imagery work is still in its early stages, but it could make the process easier for farmers submitting claims. If the directorate can establish boundaries and identify crop types from existing information, it may be able to avoid asking farmers to submit the same details again.

The potential benefit rests on turning observations into information that can be used reliably in a claim process. The interview does not describe the system’s accuracy, coverage or deployment scale, so it is too early to assess how much inspection work it might replace. Smith’s stated aim is more immediate: use the technology to reduce effort for end users, while developing its role in the service.

Changing how applications are delivered

Smith identifies application delivery as the next area to address. He wants the directorate to change how it builds and delivers applications, which he believes could give the team more room to develop new services. His stated goal is a more digital organization that can deliver services quickly and within budget.

That ambition depends on several parts moving together: choosing infrastructure according to workload, developing staff who can manage complex platforms, and designing services around the information users actually need to provide. Satellite analysis and future AI exploration may add new capabilities, but Smith’s account keeps the measure of progress grounded in delivery. The test will be whether the technology helps the directorate run its systems and makes essential processes simpler for farmers.

Neill Smith’s background

Neill Smith leads infrastructure at the Agricultural Economy Directorate. His experience includes cloud, virtualization, Linux and middleware, and he describes himself as a hands-on technical specialist as well as a team leader. He has spoken at technical events including AWS re:Invent, VMworld and Gartner Infrastructure, Operations & Cloud Strategies.