Skip to content
  • Search
  • Contact
Case Study

Digital twin customs agent reduces processing time and cost     

Overview

Customs is the gateway for all UK trade, but it still runs on manual, paper-influenced processes. Delay and error raise costs for every business that imports or exports. New requirements on border controls, safety and security data, and carbon reporting are also increasing the workload.  

MyCustoms is a Manchester-based, AI-native customs and compliance software company that automates the preparation of customs declarations and has freight operators, hauliers and customs brokers as its core customers. 

Supported by Digital Catapult and the UK Digital Twin Centre, MyCustoms worked with industry partner Emerge Manchester over seven months to create a digital twin of a customs operator.  

Automation was tested on a replica of real operations to produce a working AI agent that mirrors how an experienced customs clerk prepares, checks and corrects customs declarations. It extracts data from unstructured trade documents, validates it against live government rules and corrects errors, with human oversight and an auditable record. By the end of the project, this digital twin had become a digital colleague. 

“Customs is one of the last places where skilled people spend their days re-typing documents into forms. The Digital Twin Adoption Accelerator and the support of the UK Digital Twin Centre gave us the backing to ask a much bigger question: what if we could build a digital twin of the customs operator itself?” Luke Milne, Founder and Chief Executive Officer, MyCustoms (trading as Declaris)
7RV05998

The challenge

Customs work involves skilled staff re-keying details from unstructured documents such as invoices and packing lists into government systems, checking them against rules that change regularly, and then chasing errors. Most customs software is a form or dashboard that people fill in, with automation added on top. It is slow, expensive and error prone. 

The volume of work is set to rise for reasons including the UK Government’s planned ‘de minimis’ changes which will end the tax break on imports of goods worth less than £135, as well as the EU’s new ICS2 system for submitting safety and security declarations and carbon border requirements which will place a carbon cost on certain imported goods. 

MyCustoms wanted to go beyond digitising the data to build a dependable virtual replica of the operator’s judgement. Its approach was to create an agent that does the work, sits inside existing tools and is trained on the customer’s own data. The agent would carry out routine preparation and checking to enable the skilled staff to focus on the exceptions. Emerge Manchester, a freight operator which prepares import and export customs filings, was operating under the same pressures.  

Developing the solution

Beginning in November 2025 as an effort to digitise trade data and procedures, the project grew over the next eight months into a digital twin of the customs operator itself: a software agent that behaves like a capable member of the team. At a high level it works in three steps:  

  1. It reads the documents that arrive with a shipment and extracts the relevant data.  
  2. It checks that data against government rules as they apply at that moment.  
  3. It corrects errors itself or flags them to a person, leaving an auditable record of what was done and why.  

The agent is a specialist system trained on customs workflows, not a general-purpose chatbot, and was tested against shipment test data across four types of filing, comparing its output with declarations already filed by experienced operators. This meant the digital twin could be measured directly against how people do the job today without disrupting live operations. 

Emerge Manchester’s contribution was practical and sustained. Test data came from client jobs, with sensitive information anonymised, at a rate of tens of jobs a week. Each batch went through technical review until accuracy was above 95%, followed by a final review by Emerge’s operators. These operators took part in human-in-the-loop testing of the agent’s outputs and in modelling different scenarios, so the agent was shaped by how experienced people handle the work. 

The project developed in stages, one type of filing at a time. 

  • Across 17 test runs of entry summary declarations on 11 shipment packs, the agent’s output matched the declaration filed on the key fields, including the vehicle, route, ports of loading and unloading, and goods lines.  
  • On a live test of transit declarations, all 15 of the 15 key fields checked matched the source documents.  
  • Four tests showed that the digital twin agent correctly identified the movement type, the unaccompanied trailer and its number, the direction of travel, the crossing, and the linked declaration references on every run.  
  • For entry summary declarations, the agent’s output was compared with the declaration filed. 
“Our operators were part of the testing from the start, not just on the receiving end. They reviewed the agent’s outputs and worked through different scenarios, so we could see how it behaved before relying on it. That gave the team confidence, and it means the agent reflects the reality of how we work.” Sam Bennion, UK Commercial Manager, Emerge Manchester Ltd

Collaboration through the UK Digital Twin Centre

Customs operations are well suited to a digital twin because they are rule-based, high in volume and repeatable, yet still depends on human judgement.  

Digital Catapult’s Digital Twin Adoption Accelerator gave MyCustoms time and funding to pursue a bigger idea than it could have justified commercially and gave the work a credible platform. Without the support from Digital Catapult and the UK Digital Twin team, the work may have progressed but more slowly and with a narrower scope. MyCustoms would have prioritised near-term revenue features and the work would most likely have stayed at data digitisation rather than extending to a digital twin of the operator. 

Digital Catapult’s support sharpened the teams’ sense of what the industry needs: proven, live testbeds rather than more pilots, and a route to adoption that does not force customers to change their software. That reinforced the decision to build with a real operator, Emerge Manchester, and to design the agent to work inside the tools people already use.  

Exposure to the wider digital trade community also gave a new perspective on legislation and standards, including the legal recognition of electronic trade documents. Emerge Manchester was central to this: its operators brought day-to-day customs expertise and tested the agent’s outputs, while MyCustoms brought the capability. 

Outcomes and impact 

MyCustoms’ figures show a reduction of over 90% in both processing time and cost per declaration for standard customs declarations through the UK Customs Declaration Service. Declarations that cost £25-£50 each are now serviceable at a fraction of that cost. The technical cost to serve dropped in a similar manner and time to completion went down to less than five minutes per declaration.  

For the wider industry, the project demonstrates a route to automation that is auditable and keeps people in control, key demands from regulators and customers to prove trust in the tool.

Looking ahead 

MyCustoms’ partnership with Emerge Manchester became a live commercial opportunity with Emerge now using the software. Assuming a successful pilot, the next step will be to extend use across the wider Emerge group, with Emerge’s operators involved in a phased onboarding.  

Validation with a real operator has strengthened MyCustoms’ proposition in conversations with prospective customers and investors. There are hopes of new UK customers among freight operators, hauliers and customs brokers, where the same problems exist. There may also be expansion into EU trade lanes as ICS2 requirements grow, and development of adjacent products such as support for carbon border reporting. 

“We wanted to see whether AI could genuinely take on the routine work in customs, not just perform in a demonstration. Working with MyCustoms, first on test data and now in our own operations, has shown us a practical route, and we are looking forward to extending it across the wider Emerge group.” Sam Bennion, UK Commercial Manager, Emerge Manchester Ltd