MHCLG develops AI Gateway to simplify secure AI adoption across digital services

The Ministry of Housing, Communities and Local Government (MHCLG) has developed a centralised AI platform designed to help its digital teams adopt the technology more quickly and securely, while improving oversight of how the technology is used across the department.

The AI Gateway provides a single, governed route to approved AI services, removing the need for individual teams to develop their own integrations, security arrangements and support models.

According to a blog published by the department, the platform has already been adopted by multiple teams and provides access to approximately 20 AI models. Several initiatives are using the Gateway, with further projects progressing through onboarding.

The department said the approach is intended to reduce development costs and duplication, while allowing teams to concentrate on improving digital services rather than managing the underlying AI infrastructure.

A shared approach to AI adoption

The Gateway was developed by MHCLG’s Technical Architecture team, working with colleagues across digital, cloud, DevOps, security, governance and delivery functions.

Built using Microsoft Azure API Management, it provides a standardised API through which teams can access approved AI technologies.

The platform is designed to be AI provider-agnostic, supporting models from multiple providers through a consistent approach and allowing additional models to be integrated in future.

It can be used by developers, data scientists, applications and services running in both Azure and AWS, enabling AI capabilities to be embedded directly into departmental digital services rather than accessed solely through individual devices.

Sam Harrison, technical architect at MHCLG, explained that the department wanted to avoid a situation in which multiple teams independently developed connections to different AI services, creating duplication, delays and inconsistent standards.

Instead, the team developed a reusable platform with shared controls, common onboarding processes and a consistent support model.

The Gateway provides teams with a standardised route to AI services, reducing the development effort required to introduce AI capabilities into their projects.

Security and governance built in

A central element of the project is establishing consistent security and governance arrangements as AI becomes more widely embedded in departmental services.

The Gateway restricts access to approved models and vendors, enforces departmental security controls and supports audit and monitoring requirements.

It also provides central oversight of AI usage and is designed to promote compliance with departmental standards.

Before deployment, the platform underwent MHCLG’s internal AI approval process, alongside technical and data protection approvals and penetration testing.

It uses existing authentication and role-based access control (RBAC) methods, with integration into departmental monitoring, logging and audit systems.

Governance documentation covers acceptable use, accountability, operational responsibilities and data handling. The platform also connects to MHCLG’s wider AI governance arrangements, including its AI Register and approval process.

Harrison said the team had sought to demonstrate that AI could be adopted ambitiously and practically while maintaining appropriate levels of security, compliance and public trust.

Greater visibility of AI spending

The Gateway is also intended to give MHCLG a clearer picture of how AI is being used across its digital services.

MHCLG said its centralised approach allows the department to monitor AI consumption, identify usage and adoption patterns, and track costs and resource use.

This visibility is designed to help teams understand demand, manage spending and optimise their use of AI services.

The department said the platform has already made it easier and faster for colleagues to access approved AI technologies, with common controls, onboarding guidance, monitoring and support in place.

By reducing the need to recreate technical foundations for individual projects, MHCLG expects teams to spend less time on infrastructure and more time exploring how AI can improve services and support users.

The blog doesn’t provide quantified savings or performance measurements, although it identifies reduced development time and costs as benefits of the approach.

Further expansion planned

MHCLG intends to expand the AI capabilities available through the Gateway as demand grows.

Future development will include access to AI services hosted by additional cloud providers, giving teams greater flexibility while retaining consistent governance, security and operational standards.

The department is also inviting colleagues working on similar challenges elsewhere in government to get in touch and learn more about the project.