Want to get bold work like this done?
All case studies
Leading Third-Party Logistics Company

Using AI/ML to automate data entry across tens of thousands of daily documents

How we helped
Aim
Engage
Grow

Computer vision to cut through the complex

A leading third-party logistics (3PL) company was using scores of data entry operators to manually enter bill of lading information from 50,000+ scanned documents every day. These bills have hundreds of layouts, which made prior automation efforts using tools such as Optical Character Recognition, commonly referred to as OCR, fail. The client engaged Launch by NTT DATA to accelerate their build of a reimagined Bill Entry product with technical innovations and a stretch goal of exploring use of computer vision and AI/ML capabilities to improve biller experience and elevate their efforts.

The ask
No items found.

Atomizing the work while building the business case

Launch’s Al team started by framing the business problem into small, discrete parts that could be individually engineered and tested. This approach allowed for the level of technical scrutiny required to research and test potential solutions effectively and individually. Simultaneously, the team collected historical cost and KPIs to gauge success and build the business case for implementing Al bill entry automation. The technical roadmap and business value were both clearly demonstrated, paving the way to move forward with confidence.

An MVP that proved it

Within a few months' time, Launch by NTT DATA trained and tested multiple, custom-built machine learning models paired with the most effective third-party AI services to generate a business-viable solution recommendation that excited the client. With a concrete business case and a technical MVP prove-out, it is now time to scale the solution.

Power up the AI smarts

Launch by NTT DATA collected additional ML training examples and incorporated human feedback back into the Al training process. We expanded scope to include additional data elements, and managed the experimentation to MVP pipeline for each additional element. This successful outcome sets up the solution for future scaling and operationalization.

The impact

This solution is projected to save the client up to 550 data entry hours per night through automated data extraction and bill entry across 50K+ bill of lading documents daily.

,

Download the app