Back in 2023, the UPU’s State of the Postal Sector Report identified artificial intelligence (AI) as a “beacon of hope” for the international postal industry. The report explored how the technology can be used for more than just automation and efficiency gains. AI presents an opportunity for hyper-collaboration between multiple ecosystemic stakeholders, ushering in a new era of sustainable postal development and helping to rectify the imbalances in both national and international postal development.
“International postal services can only succeed with AI if they intensify data sharing between them and with all their transportation and cross-border e-commerce partners,” said UPU economist José Anson. “We call this “hyper-collaboration” in the international postal sector.”The transformation potential of AI has not gone unnoticed by the postal industry, with many operators around the world looking at how they can embrace the technology for optimized routing, real-time tracking, and personalized services.
“AI will increasingly support the establishment of digital twins throughout the entire postal supply chain and processes,” Anson continued. This will become a game-changer in terms of increased efficiencies in a time when postal services and other logistics operators struggle to maintain their profit margins.”
Anson believes that “an agile mindset and organization” is essential when seeking to capitalize on AI. He also stresses that posts “must act now” when it comes to AI and that they should test “as many use cases as possible” to find the best ones.
The UPU has been working on several AI innovations in recent years, developed to support its members and their customers. These include the use of AI in its Global Track and Trace (GTT) system, AI to detect dangerous and counterfeit goods, and for stamp identification.
Global Track and Trace
The GTT system, which provides international routing information of postal items for operators and e-commerce platforms worldwide, now includes a prediction engine that uses machine learning models trained on postal big data to predict estimated delivery days.
“UPU GTT is used in many countries all around the world,” said Silviu Lita, Software Architect at the UPU. “Currently, it gets about 1-2 million hits per day.” The system has been extremely successful over the years, however, the provision of reliable information on the expected delivery date of a postal item has been a challenge, especially for postal operators in the least developing, developing countries and small island developing states.
But now, using years of postal big data, the new engine can estimate delivery days with surprising accuracy. “It uses our big data environment to train a machine learning model,” said Lita. “Then it uses the trained model to predict the delivery. The prediction is re-evaluated as new events are captured for a mail item.”
Talking about the accuracy of the new tool, Lita continues, “The prediction accuracy varies. It’s lower at the time the item is posted – about 61% – but it improves significantly, reaching 91% once the item is released from inbound customs.”
Developed entirely in-house, the AI module (Global Track and Trace system) represents a major technological milestone for the UPU. Posts can either embed the GTT interface into their websites or access the prediction engine through an API.
Dangerous Goods Search Tool
The UPU is also embracing AI in the development of its new Dangerous Goods Search Tool (DGST), which is an AI-powered system that can detect high-risk items in the global postal system.
“The DGST is currently undergoing structured User Acceptance Testing (UAT) with 40 designated postal operators worldwide,” said Dawn Wilkes, Security Program Manager at the UPU. “This phase is critical – it allows us to test how well the AI identifies dangerous goods using real data from international mail events and customs declarations.”
The DGST calculates a Dangerous Goods Prediction Score (DGPS) for each item, using machine learning models trained on vast postal datasets. “It processes EDI messages like ITMATT and EMSEVT in near real time and returns a score through an API,” Wilkes explained. “Operators can then take appropriate action based on that risk assessment.”
The tool will also become smarter over time. “We will retrain the model regularly – every three months or so – to improve its accuracy,” Wilkes added. “The more data it ingests, the more intelligent it becomes. That’s why we encourage posts to authorize data access – it’s a win-win.”
But the UPU isn’t stopping there. “One of our key goals is to integrate DGST with external intelligence sources,” said Wilkes. “We’re exploring secure data-sharing partnerships with customs, UN agencies, and national enforcement bodies to cross-reference flagged items with known threats, such as blacklisted senders or high-risk trade routes.”
Set for broader rollout by late 2025, DGST promises to revolutionize mail security. “Ultimately, it gives posts a powerful tool to protect employees, cut costs, and ensure compliance,” Wilkes concluded. “This is about future-proofing the global postal network.”
Stamp identification and counterfeit usage
Security is also at the heart of the UPU’s World Association for the Development of Philately Numbering System (WNS) app, which functions as a mobile extension of the WNS website – UPU’s authoritative stamp database since its inception in 2002. Dubbed WNS 2.0, the AI-powered mobile app enables instant stamp identification and combats global counterfeiting.
“This app is a game-changer for the philatelic community,” said Benjamin Combes, Manager of the UPU’s Philately Development and Documentary Heritage Programme. “It offers customs and law enforcement a new tool to fight counterfeiting, encourage in rebuilding trust between posts and collectors, and make stamp information easily accessible to everyone.”
The WNS app draws from an early database of more than 120,000 officially verified stamps for now. Using AI, users simply take a photo of a stamp to access detailed metadata in seconds, replacing magnifying glasses and catalogs with a smartphone. “With the WNS app, users can identify a stamp instantly. For collectors, it’s an incredible leap forward,” Combes said.
Beyond collectors, posts stand to benefit significantly. “Counterfeiting costs the postal world millions,” Combes explained. “The WNS app helps verify authenticity by cross-checking stamps against our trusted database.” Therefore, customs agents can use the app to inspect stamps without repeatedly contacting postal authorities, streamlining border checks.
Importantly, WNS 2.0 brings stamp collecting into the digital age. “If it’s not on their phone, it doesn’t exist for the younger generation,” Combes said. Future updates will include social sharing, real-time market values, and broader language support. “WNS 2.0 is just the beginning,” Combes concluded. “We're building a smart, global philatelic platform for the future.”
The future of AI
While the UPU continues to work on groundbreaking new AI-based tools to support its members, and encourage hyper-collaboration, Anson believes that there are still many underexplored areas when it comes to AI.
He concluded: “For example, agentic AI – the kind of AI that moves organizations from automated to autonomous processes and optimizes operations, will be the next wave of AI innovation in the postal sector and will show the real value proposition of AI for postal companies.”
This article first appeared in Union Postale Summer 2025