Everyday AI for freight forwarders

October 18, 2022
Raft

Share this article:

During a recent webinar with our customer Adam Hill from Scarbrough and our partner Shane D'Aprile from OLC Consulting, we discussed many use cases and examples of every day AI usage by freight forwarders. Th audience posed many questions for live answers. We've recorded the questions & answers below:

Questions & Answers from the live audience:

General Features

Can the platform read scanned PDFs, JPEG as well as text-based PDFs?

Yes. In addition to other files formats such as .xls, CSV, EDI, etc.

Does your platform work with TMSes other than CargoWise?

Yes, we integrate with multiple TMS and ERP systems. While CargoWise is one of our major integrations, we also support many other major TMSes along with custom in-house TMSes. We also have a generic API that can be used to build a custom integration and a bunch of tools to configure the output from our system into any xml, json or EDI format - allowing new integrations to be built very quickly.

For your emissions product, can this data be forwarded directly to end customers?

Yes, a core part of our emissions function calculates emissions data for every leg of the shipment and every mode of transport. These calculations are then added back into CargoWise for regulatory/compliance/reporting reasons and to make the data available for your customers in dashboards and reports.

How do you deal with privacy of data?

All customer data is segregated by default - whether in our databases or in cloud storage buckets. We have strong data security protection and IP protection policies in place and are working towards ISO 27001 and SOC2 compliance.

CargoWise

How does your AP matching work? Can you discuss how invoice descriptions are matched to CargoWise charge codes?

The first level of validation between invoice and accruals uses the creditor code, invoice line amount, currency and tax. Where necessary, we also validate using the charge code in accruals and the description on invoice. This works based on feedback provided by a user who can specify the charge code that a description should match against for a vendor, which will then automatically be used for subsequent invoices.

Is the platform smart enough to apportion consol costs down to the shipment?

Yes, we handle charges across shipments, consols, brokerage files, warehouse files, etc. For consols specifically, we can match and post consol level costs, gateway level costs and also apportioned sub-shipment costs.

What is the success of airfreight shipment registrations as they tend to be very urgent?

Our pre-alert registration product supports air freight as well as sea freight. The automated job creation should significantly reduce the time it takes for a user to create a job, likely reducing the need for work overflowing into weekends and public holidays.

Can you integrate with other modules in CargoWise like warehouse, transport, etc. ?

Yes, we integrate with multiple modules in CargoWise, both accounting and operational, using our flexible integration framework. We have a dedicated integrations team that has a lot of experience with CargoWise, allowing us to build and maintain new integration points with CargoWise quickly.

Machine Learning

Do you indicate confidence levels of the extraction to guide a user to true exceptions?

We validate all the data that we can in order to bring to the user's attention potential discrepancies, issues and low confidence data to allow them to drive by exception. We validate data at multiple levels - we compare financial information and accruals, master data records, shipment information from the TMS, information between documents, etc. We also carry out mathematical checks, checks for mandatory data, etc.

Can all learning for common document types be incorporated at the start since HBLs, MBLs, AP are common for all freight forwarders?

Yes, by default our AI technology has two combined components: 1. General models that have been trained on multiple document types, multiple forwarders and thousands of examples. 2. Customer specific feedback that fine tunes the models on their specific document distribution and preferences.

If a PDF has multiple documents, how does it work? Does it have the intelligence to separate them out?

Yes, the first step in our processing pipeline is to separate and categorize documents whether they are all bundled into a single PDF or across multiple attachments. We support all kinds of attachments - PDF, JPEG, TIFF, EXCEL, CSV, EDI, etc. Once the documents have been categorized correctly, the correct data fields are extracted from all the documents.

How does the feedback and learning component of the solution work?

TBD

Roadmap

Visibility is interesting part of your system to complement other functionality already present.

TBD

I wonder if you can bring in other data sources for further data validation? For example, data from port terminals to validate voyage numbers?

As a platform, bring additional sources of data in is very valuable to us and a core part of the vision. Whether its data from a TMS, a document, a visibility provider, a port terminal or anything else - making it accessible in a standardized format is a very important aspect of the roadmap.

Is auto-costing on the roadmap? Getting rates data from carrier rate spreadsheets would save a lot of time!

Yes, we plan to integrate with CargoWise rate and costings! We’re actively working on it right now.

Become more than
a freight forwarding partner.

Discover just how indispensable you can be.

By clicking “Accept All Cookies”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Policy for more information.