CustomAI has reached a pivotal milestone in its mission to revolutionize European border security. Project partner ISDEFE leading the “End-User Requirements” task has officially circulated a series of comprehensive use-case questionnaires to project partners, signaling the start of the Operational Co-Design & Integration phase (Work Package 3).
Over the next month, CustomAI partners will provide the “Ground Truth” data required to transform AI from a theoretical concept into a practical tool for customs officers at sea, in the air, and in the lab.
Closing the “Gap”: From Data to Decisions
CustomAI is tackling a massive challenge: increasing the inspection “hit rate” (detection success) from a 1% baseline to a target of 10%, all while increasing positive interventions without expanding human resources. To achieve this, the project is developing a Virtual Customs Control Office (VCCO) that fuses data from Non-Intrusive Inspection (NII) scanners, novel vapor sensors (MION), and blockchain traceability.
“We aren’t building ‘black box’ algorithms,” noted the ISDEFE team behind this activity. “These questionnaires are to ensure our AI is grounded in the physical reality of the field. We are currently in the ‘Gap Analysis’ phase, in which we are defining how an officer identifies a threat so that then the AI can replicate and enhance that expertise.”
Three Use Cases: Operational Co-Design
The ongoing data collection targets three distinct environments simultaneously, ensuring that the AI models are tailored to specific logistical pressures:
UC1: High-Volume Maritime Container Inspections: As the designated Pilot Hub for large-scale cargo, the Port of Valencia serves as the primary ground for testing these technologies before they are scaled to ports in Constanta and Aarhus. This use case focuses on high-volume container traffic, aiming to detect complex threats like the “Rip-off” method and concealments in refrigerated units. In this phase, partners are providing the “Ground Truth” data, such as defining how officers visually identify engine-bay concealment, to ensure the AI minimizes “false positives” that disrupt legitimate logistics chains.
UC2: High-Frequency Parcel Inspections: As the designated pilot for high-frequency logistics, the Copenhagen Airport (CPH) serves as the testing ground for the “torrent of needles” inherent in postal and courier services. In collaboration with the Danish Customs Agency (TOLD), this pilot aims to integrate AI-driven “Stop/Go” decision support directly into high-speed sorting belts. Partners’ responses will define the millisecond latency required for automated diversion, ensuring the system enhances human decision-making in high-pressure e-commerce environments.
UC3: Detection of Illegally Exported Cultural Goods: Led by DGGAM and MOT, this specialized lab demonstration focuses on high-complexity interdiction. Unlike high-volume cargo, this pilot targets sophisticated concealment methods, such as the “Marble-in-Marble” technique. Partners are currently defining the visual and structural features that distinguish genuine artifacts from replicas, helping to train models that provide forensic-level accuracy through 3D image database querying.
The Questionnaire Structure
To ensure a holistic view of the operational landscape, each questionnaire follows a five-part structure:
- Part 0 & 1: Respondent Profiling and Digital Maturity (Gap Analysis).
- Part 2 & 3: Strategic Definition and Operational Co-design (Pilot/Lab Scenarios).
- Part 4: IT Landscape & Data (Legacy vs. Future Integration).
- Part 5: Legal, Ethical, and Privacy Constraints.
The insights gathered over the next month will set the operational thresholds for the entire CustomAI ecosystem, moving the project closer to its goal of dynamic, AI-driven predictive profiling for a more secure Europe.
