HTX partners with Mistral AI and Microsoft to build advanced AI models for Singapore Home Team

HTX Partners with Mistral AI and Microsoft to Accelerate AI Model Development for Home Team

May 26, 2025 | Singapore News Center

HTX (Home Team Science and Technology Agency) has signed a contract with Mistral AI and Microsoft to enhance its ability to develop advanced Generative AI models and platforms. These models aim to improve operational efficiency by automating routine tasks and generating actionable insights for the Home Team.

Since its establishment, HTX has focused on leveraging technology to support Home Team operations. Last year, the agency initiated an AI movement to boost operational efficiency, allowing officers to concentrate on critical safety and security roles that directly benefit the public.

Ang Chee Wee, HTX Chief AI Officer and Assistant Chief Executive (Digital and Enterprise), highlighted the significance of this collaboration: “Our goal is to use AI to improve operational efficiency, enabling Home Team officers to focus on more critical areas. Partnering with Mistral AI and Microsoft will speed up our AI development efforts.”

Collaboration Details

The partnership designates Mistral AI and Microsoft as HTX's research collaborators. Microsoft will assist in fine-tuning Phi-4-multimodal, a Small Language Model (SLM) tailored for Home Team applications. Additionally, Microsoft will help integrate SLM training frameworks into HTX’s AI platform and infrastructure.

Mistral AI will support the fine-tuning of their Large Language Models (LLMs) for Home Team use and collaborate with HTX to develop a proprietary series of LLMs named Phoenix. Mistral AI brings expertise in LLM training, while Microsoft anchors the technology platform and infrastructure.

Wong Sook Huey, Public Sector Director at Microsoft Singapore, emphasized the value of this public-private partnership: “By combining HTX’s mission with our technology and Mistral AI’s expertise, we’re building local AI capabilities that meet the specific safety needs of the Singapore community.”

About Phoenix: The Home Team’s Custom AI Model

Phoenix has been pre-trained on extensive datasets relevant to the Home Team and Singapore, including support for 10 major local languages such as Mandarin, Bahasa Melayu, and Tamil. This specialized training allows Phoenix to deliver more accurate responses compared to general-purpose LLMs.

The model will be deployed progressively as an AI chatbot for Home Team officers and as an API for developers to build operational applications. This initiative is a major step toward establishing sovereign AI capabilities within the Home Team.

Arthur Mensch, CEO and co-founder of Mistral AI, stated, “We are excited to co-train models like Phoenix with HTX. This collaboration strengthens our commitment to the APAC region and advances AI tailored to government and public safety needs.”

Building a Unified AI Platform

The contract includes deploying Mistral AI’s platform to enable inference, fine-tuning of APIs, and custom model building for various Home Team departments. HTX will develop a unified platform to host its LLMs and Vision Language Models (VLMs), simplifying AI access across the agency.

This unified platform will democratize AI use by making state-of-the-art models readily accessible to Home Team officers and developers. It also positions HTX as a leader in AI development, as few organizations in Singapore have built and deployed their own LLMs from the ground up.

Implications for IT and Development Professionals

• Expect increased demand for expertise in AI model fine-tuning, deployment, and infrastructure integration within government-linked agencies.
• The development of Phoenix and the unified AI platform offers new opportunities to build specialized AI applications that improve operational workflows.
• Collaboration between public agencies and AI companies like Mistral AI and Microsoft highlights the importance of cross-sector partnerships in AI innovation.

For IT and operations professionals aiming to deepen their AI knowledge and skills, exploring relevant courses on AI model development and deployment can provide practical advantages. Resources such as Complete AI Training’s latest AI courses offer focused learning paths tailored to these emerging needs.