TranSafeAI @ IATA One Record Hackathon by ICCS

Cover Page
High-Level Flow
Sample API-Service

Inspiration

The rapid growth of ecommerce presents a significant challenge for shippers, freight forwarders, and carriers in ensuring the secure transportation of high volumes of shipments. In particular, the risk of dangerous goods being loaded onto aircraft due to x-ray feasibility and human error, as well as hidden dangerous goods due to fraudulent declarations and human mistakes, poses a serious threat to the shipment process safety, revenue, and lead time.

Inspired by the latest ICS2 regulations requiring shippers to provide detailed information on shipments to the EU, we propose leveraging AI models to map Harmonized Codes with UN numbers. By developing two API services with 1R standards at the House Airway Bill (HAWB) and Master Airway Bill (MAWB) levels, we aim to enhance the ability of freight forwarders and carriers to identify and flag any dangerous goods and embargo goods in ecommerce shipments. This will ensure the safety of the shipment process and personnel involved, as well as minimize monetary losses for carriers and freight forwarders. We aim to tackle the below challenges: • eCommerce Challenge: Develop a solution to e-commerce & air cargo, for example in the domain of customs & border, detecting undeclared dangerous goods, Customer Experience, Package Tracking, screening etc. • AI Challenge : Create an AI-powered tool or system that leverages IATA's ONE Record standard to enhance air cargo processes, optimization, efficiency, prediction detection etc. • CHAMP: How can we leverage the CHAMP APIs for innovative use cases around eCommerce, sustainability and ONE Record.

What it does

By providing two API-services in 1R standard that connect to AI-powered databases of HS codes, UN codes, and embargo policies, we aim to improve safety and compliance in the supply chain. The first API-service enables shippers and freight forwarders to easily identify DGs at the HAWB level, ensuring proper validation and checks are in place early on. The second API-service allows carriers to identify DGs at the MAWB level and cancel bookings through the CHAMP API, providing visibility on DGs to prevent any hidden or undetectable dangerous goods from reaching the cargo build-up stage. By facilitating early detection and communication of DGs, our solution helps enhance safety measures and compliance throughout the supply chain.

How we built it

We utilize the Llama2 model, a collection of pre-trained and fine-tuned large language models (LLMs) with parameter sizes ranging from 7 billion to 70 billion, to accurately map HS Code and Product Description, enabling us to extract UN Numbers and embargo policies. This helps us prevent the shipment of undeclared and undetectable dangerous goods in all ecommerce shipments. Rather than analyzing each HAWB individually with AI, we have implemented a standalone program that uses AI to analyze HS Codes, UN Codes, and embargo policies and update the databases. This allows the API services to quickly index data, even when enquiring more than thousands of HAWB data within a single MAWB.

Challenges we ran into

• Time running so fast. • It is challenging to put all our ideas in 5 mins video

Accomplishments that we're proud of

• Having a supportive, open-minded and passionate team • While applying AI to address the business problem, we also utilize AI in assisting us in generating codes, writing scripts and video editing. Allowing us to focus on tackling the business problem with an ideal solution.

What we learned

• Think slow, act fast – carefully consider all aspects of a situation and stakeholders’ benefits before acting, leading to better outcomes in the long run. • While technology unlash a lot of possibilities to many business problem, delivering a clean and standard solutions to address the real problem also crucial in driving digitalization.

What's next for TranSafeAI @ IATA One Record Hackathon by ICCS

• Testing and refining the AI model to ensure accurate mapping of HS codes, UN numbers, and embargo policies. • Developing a user-friendly interface for shippers, freight forwarders, and carriers to easily access and utilize the API services. • Continuously monitoring and updating the databases to ensure they remain up-to-date with the latest regulations and requirements related to dangerous goods transportation.