UPU

Universal Postal Union

9
Activities on AI
5
Countries

About UPU

One world. One postal network. The Universal Postal Union is a United Nations specialized agency and the postal sector's primary forum for international cooperation.

Activities by UPU

UPU
UPU Unified Data Platform and Agentic AI for Postal Development
Global Not Applicable

agent.udp.post is a UPU initiative to build a secure, trusted and ethical agentic AI ecosystem for the postal sector, powered by the UPU Unified Data Platform and hosted within the UPU’s trusted .POST digital environment. The initiative brings together postal and non-postal data, including UPU big data and relevant sources from trade, logistics and aviation, to enable AI agents that can analyze postal development, diagnose country-level challenges and advise on policy, regulatory and operational improvements.
The first pilot agent focuses on postal development, helping member countries identify gaps in reach, relevance, reliability and resilience, and test potential solutions using data-driven insights. Over time, the initiative aims to evolve into a multi-agent ecosystem, where specialized agents support areas such as postal development diagnostics, strategic planning, consensus-building, interoperability, customs clearance, route optimization and last-mile delivery.
As a UN specialized agency, the UPU positions agent.udp.post as a collective intelligence infrastructure for the global postal network. The system applies granular data governance, rights and permissions so that member countries retain control over whether and how their data is shared across the network.

Activity Type AI Tools/SolutionsPolicy/Regulatory Guidance
UPU
Prediction of postal item delivery day on the UPU global track and trace system
Global

The UPU’s global postal supply chain consists of almost 700,000 postal access points and reaches 95% of the world’s population for the delivery of postal items; mail, parcels and postal payments. In the delivery process to the addressee, the provision of reliable information on the expected delivery date of the postal item is a challenge, especially so for postal operations in least developing, developing countries and small island developing states. Using machine learning models trained on postal big data, the UPU’s global track and trace solution (

Activity Type AI Tools/Solutions
UPU
UPU Unified Data Platform and Agentic AI for Postal Development
Global Not Applicable

agent.udp.post is a UPU initiative to build a secure, trusted and ethical agentic AI ecosystem for the postal sector, powered by the UPU Unified Data Platform and hosted within the UPU’s trusted .POST digital environment. The initiative brings together postal and non-postal data, including UPU big data and relevant sources from trade, logistics and aviation, to enable AI agents that can analyze postal development, diagnose country-level challenges and advise on policy, regulatory and operational improvements.
The first pilot agent focuses on postal development, helping member countries identify gaps in reach, relevance, reliability and resilience, and test potential solutions using data-driven insights. Over time, the initiative aims to evolve into a multi-agent ecosystem, where specialized agents support areas such as postal development diagnostics, strategic planning, consensus-building, interoperability, customs clearance, route optimization and last-mile delivery.
As a neutral UN specialized agency, the UPU positions agent.udp.post as a collective intelligence infrastructure for the global postal network, not as a commercial software platform. The system applies granular data governance, rights and permissions so that member countries retain control over whether and how their data is shared across the network.

Activity Type AI Tools/Solutions
UPU
Detection of dangerous (and prohibited) goods in the UPU postal network
Global Not Applicable

The UPU postal network is sometimes used for illegal traffic (such as: drugs, counterfeit items) and dangerous items (such as: inflammable/explosive products) that may get injected into the network. The Dangerous Goods Search tool (DGST) uses machine learning models trained on postal big data, to score the likelihood of the item to be dangerous (or prohibited). At the time of this report, the system is still under User Acceptance Tests (UAT) and we do not have as yet metrics to report. In the future, there are plans to interface the DGST with other intelligence databases, and offer a comprehensive tool to postal officers in charge of protecting the postal network.

Activity Type AI Tools/Solutions