ZeroOneFour - One Chat

• 

  ZeroOneFour Team Photo

• 

  ZeroOneFour Team Photo

• 

  e-commerce store front

• 

  e-commerce selection

• 

  e-commerce fulfilment

• 

  CMS platform to send updates from

• 

  CMS platform view DG declaration

• 

  CMS platform flight finalization

Inspiration

Our inspiration for our project 1NE Chat was the need to streamline information sharing. Using AI to identify and validate DG, Project 1NE Chat maintains the flow of e-commerce goods end to end. An additional benefit is the ability to link rates, reducing emails to partner airlines, which is an extremely manual process. Customers benefit from realtime milestone updates - including the unhappy path resolution.

What it does

Our solution is a comprehensive modern system that streamlines complex logistics processes. Leveraging AI, it generates accurate descriptions for airlines to minimize the risk of undeclared dangerous goods. Precise descriptions for customs authorities, expediting clearance at the destination Shares logistics milestones with third-party providers, e-commerce stores, end consumers, and customs authorities Streamline complex interline rates processes Sustainably, it reduces the reliance of paper documentation

How we built it

We utilized a locally deployed Mistral 7B LLM utilizing an OpenAI compliant API on Apple Silicon to validate the dangerous goods. In a real-world scenario the LLM can be further trained for precision. To aid with precision we utilized retrieval augmented generation (RAG), where we are using a reference database for descriptions and to retrieve shipper confidence scores.

Finally, as a bonus, end consumers can receive updates through WhatsApp, as our platform utilizes a message broker. The broker is platform agnostic and can be linked to other well-known messaging applications.

Challenges we ran into

Our initial idea was to use AI to figure out product description and undeclared dangerous goods. For the generated descriptions, the model was unfortunately insufficient and scored low in our tests. We pivoted to undeclared DG using SKUs from the e-commerce site and had AI validate descriptions using that data source.

Accomplishments that we're proud of

We are proud of the AI validating dangerous good Sending notifications to end consumers via WhatsApp Setting up an airline CMS utilizing CHAMP APIs and updating OneRecord

What we learned

Not only did we learn how integral logistics communication is end-to end, with respect to eCommerce (shipper to end consumer), but also how to better streamline this communication, both, through the happy and unhappy path. We also gained valuable knowledge surrounding DG, and the impacts undeclared goods have on the eCommerce logistics chain.

What's next for ZeroOneFour

Continuing to leverage AI and augmented reality to better streamline cargo operational processes, further strengthening our solutions in the future.

Built With

• angular.js
• asp.net
• c#
• divinci-resolve
• docker
• lmstudio
• postman
• python
• rabbitmq
• swaggerui
• typescript