Building a security cleared AI capability at the FCDO

The FCDO partnered with Ten10 to implement a hybrid consultancy model that builds sovereign AI capability rather than vendor dependency. By deploying a blended squad of senior architects alongside Ten10 Academy Consultants who were vetted and trained in the FCDO’s specific tech stack before arriving on-site, we helped establish the ‘Agent Factory’. This repeatable operating model ensures every solution meets high standards and can be developed consistently, as detailed in the official FCDO case study below:

The Foreign, Commonwealth and Development Office works in one of the most sensitive environments in government. Much of its work involves information that requires Developed Vetting as standard. At the same time, demand for AI skills across the economy has grown sharply. Bringing these two pressures together created a clear challenge for the FCDO. The department needed people with strong technical skills who could operate safely in a high security setting, but those people were in short supply across the whole country.

Interest in AI inside the organisation was rising quickly. Staff across the network could see opportunities for AI to reduce manual effort, support decisions and unlock time for higher value work. The ideas were grounded in real operational needs. The question was how to move from interest to responsible adoption in a controlled and secure way. AI is still an emerging capability in government and needs are not always clear at the start. Requirements evolve through learning and experimentation. The FCDO therefore needed a model flexible enough to develop new ideas while robust enough to meet strict security and governance standards.

Traditional options could not solve this on their own. Recruiting permanent staff with both AI expertise and DV clearance was extremely difficult. Buying a complete ready built service risked creating a dependency that would limit the department’s long term capability. The FCDO needed an approach that combined expertise, collaboration and skills development. It needed a model that grew capability inside the department, not outside it.

The first step was to create a single, simple route for AI ideas to enter the system. Instead of separate pilots developing in isolation, all proposals flowed through one point where they could be explored, understood and prioritised. This gave the organisation a clear picture of demand and highlighted where different parts of the department were facing similar challenges. It also helped senior leaders make informed decisions based on a complete view, rather than scattered requests.

Alongside this, the FCDO developed a unified way of exploring and delivering AI work. Every idea brought questions about safety, security, data protection, ethics and value. Without a structured method, pilots could stall, drift or introduce risks that were not visible. The department therefore created a consistent approach to assessing problems, shaping solutions and providing the right safeguards. This helped build trust in how AI was being handled and ensured that work progressed responsibly from the start.

A key feature of the model was forward deployment. Instead of building solutions at a distance, the department placed technical teams directly into business areas. They worked side by side with colleagues who understood the problems best. They observed how work was done, listened to what teams needed and tested ideas in real operational settings. This way of working created strong relationships and kept the focus on practical value rather than theoretical potential. It also allowed faster learning, earlier adjustments and better adoption.

To support this, the FCDO created what became known as the Agent Factory. This is a repeatable operating model for exploring, designing and building AI tools safely. It provides patterns and guidance that ensure every solution meets high standards and can be developed consistently. Once an approach works in one part of the organisation, it can be used again for similar challenges elsewhere. This reduces duplication, shortens development time and strengthens confidence that risk is managed well.

The FCDO also recognised that long term capability depends on people. The programme placed strong emphasis on developing a diverse and talented early career pipeline. These individuals were trained in relevant technologies and given meaningful project experience from the start. Working closely with experienced leads and civil servants, they developed confidence in both the technology and the environment in which it operates. Skills were shared naturally through daily collaboration rather than through one off handovers. This helped capability grow inside the department, ready for the future.

The impact of this approach is visible in the growing pipeline of AI work. The department now has a clear route for taking ideas forward, a repeatable way of developing them and a workforce that is learning through real delivery. Staff across the organisation feel more confident exploring where AI might help them because there is a safe, consistent model sitting behind it. Leaders have better visibility of demand and can plan resources more effectively. The model has also attracted positive external recognition, reflecting the responsible and steady progress made in a complex, high security environment.

Most importantly, the FCDO is building capability that will endure. The purpose was never just to create individual tools. It was to build the people, processes and ways of working needed for long term, responsible use of AI. The department now has a foundation that can grow and adapt as the technology evolves. It shows that even in the most sensitive parts of government, it is possible to adopt AI safely while strengthening the internal workforce that will support it over time.

By implementing this model, Ten10 helped the FCDO move beyond short-term experiments to build a sovereign capability that will serve the department for years to come. This partnership ensures that, as the technology evolves, the FCDO retains a project-hardened, in-house workforce ready to run and scale these services independently.