The challenge

Adopting a long-term planning vision, the dispatcher highlighted several goals to be achieved. However, in the first instance, the challenge focused on solving some of the needs that emerged. Within this framework, the freight forwarder needed to reduce the time spent on back-office activities to:

• Create Bill of Lading (BoL) instructions from data automatically extracted from letters of credit, invoices, and customer instructions received via email;
• Automatically extract data contained in VGMs (Verified Gross Mass) received via email;
• Automatically extract data contained in Manifests (Ship's Manifest) from third-party partners received via email.

Our solution

Wenda deployed its proprietary AI solution that can automate emails and documents in daily logistics, transportation and supply chain activities.
This enabled it to proceed with the digitization, optimization and automation of the process of managing instructions for Bills of Lading, Manifests and VGMs through proprietary AI algorithms.

There are three different models for each type of document, namely instructions for Bill of Lading, Manifests and VGMs.
Wenda's AI models automatically identify and extract relevant data from the header and detail of the document, the subject and text of the email, and the files attached to the email. Specifically, the following data are extracted from each document: 

• Instructions for the Bill of Lading (BoL): shipper, consignee, notified party, reservation number, place of receipt, ocean vessel, port of embarkation, port of discharge, final destination, freight payable, package number/type, commodity description, gross and net weight;
• Cargo Manifests: container type number, seal, number of packages, cargo weight and tare weight;
• VGM: gross mass weight, container code, container type, authorized person, signer and shipper.

The use of Artificial Intelligence, unlike OCR and similar technologies, allows for understanding the context of documents and/or text. In this way, many different layouts in different documents can be processed without human intervention.

For example, if in a given Load Manifest the Container Number is located in the top right of the page and is encoded as "container number," in another Manifest it might be located in the middle of the page instead and might be encoded as "number of container."
Wenda's AI will be able to figure out that this is the same value.

This capability also eliminates the need to use a specific format such as pdf, jpeg, xml or word, because the AI will be able to read documents regardless of format.
The same flexibility exists for the creation of output files, which can be created in any format and received by the user's preferred/selected method.

Workflow

1. The client sends the forwarder an email with an attached document.
   In the case of BoL instructions, the shipper receives invoices, letters of credit, or simple instructions via email. In this case, Wenda uses a classifier AI model that can analyze and distinguish the type of document received, and then shares it to the specific AI model (invoices, letters of credit, customer instructions) that will extrapolate the data.
2. Wenda then takes over the emails received from the shipper, extracts the data from the attachments that are used to create the Bill of Lading Instructions, and sends an email to the shipper's back-office.
3. This email sent by Wenda to the shipper's back-office contains text in which the sender of the customer's email and the attached file name are made explicit.
   The attachments to this email sent by Wenda include both the original attached file and an excel file: the latter is the end result of Wenda's automated processing.

The same process is repeated for Manifests and VGMs.
In short: Wenda takes in the emails received from the shipper, automatically parses the data within each Manifest and each VGM shared by the referring partner, extracts the data contained therein, and structures it into an excel file that is then sent to the shipper's back-office.

This way, by receiving Wenda's automated emails, the shipper's back-office has a tidy file to import into its system with all the necessary information related to each document to be processed.
They no longer have to go looking for information in different emails, received at different times and from different senders: based on the document to be processed, the shipper's back-office opens the excel file received automatically from Wenda, takes the information he needs and imports it into his IT systems.

Future integrations

The ultimate goal of the customer and Wenda is total automation of back-office work to support the operators.

In order to further digitalize and optimize the shipper's back-office processes, Wenda will provide some advanced features of its proprietary AI.

These advanced features will allow Wenda's AI to be integrated with the customer's internal systems, initiating an even more efficient automation process: excel files received by the freight forwarder will be automatically entered into its ERP, further cutting document processing time and dramatically increasing back-office productivity.

In addition, the advanced capabilities of Wenda's proprietary AI will be able to enable the shipper to automatically issue shipment quotes (by dialoguing with the ERP) in response to specific quote requests received from customers via email.

Benefits

The implementation of the solution has brought remarkable results:

• 80% savings in back-office time: by reducing the time spent on document management, the shipper's staff can now focus on more value-added, more strategic, and more creative tasks;
• More efficient allocation of human resources: the freight forwarder was able to automate the processing of thousands of documents, greatly increased the productivity of human resources, and distributed workloads more efficiently;
• Greater scalability in customer service at the operational level: automation allowed the forwarder to expand its operations. Looking ahead, this element can be further enhanced by the potential of automation.