InfoCert Enhances Its ML Capabilities Using Generative AI on AWS

Previously, InfoCert built custom ML models to help classify and process emails and attachments for each Legalmail customer. Training and deploying these models was a lengthy, costly process, which limited access to this set of capabilities to larger companies. While InfoCert was searching for a way to streamline such ML integrations, AWS invited the company to preview Amazon Bedrock, a fully managed service that can be used to build and scale generative AI applications with foundation models.

“Using Amazon Bedrock, we have the opportunity to access a broad variety of pretrained models, which were not available with other vendors,” says Fabrizio Leoni, Head of Innovation Portfolio at InfoCert. “We aim to move from a custom approach to a more general solution that requires minimal adjustments and no model training.”

InfoCert conducted a proof of concept to assess the feasibility and potential of generative AI and Amazon Bedrock. Through this project, the company evolved its existing Intelligent Multichannel Classification and Routing (IMCR) system, which helps large enterprises reduce operational overhead in document and email processing. IMCR does this by automating the routing of incoming email messages and integrating with other services and tools. The new solution, IMCR AI, uses generative AI to offer a unified workflow suitable for customers with small-to-moderate information extraction needs. “By combining Amazon Bedrock’s generative models with prompt engineering, we get preliminary results that are very close to what we achieved with the previous solution, which required months for data preparation, model training, and rules definition,” says Leoni.

Using Amazon Bedrock, InfoCert can now integrate ML models faster and more efficiently; it anticipates a decrease in the time required for model definition, training, and refining measurable in terms of weeks instead of months. Additionally, InfoCert expects to reduce financial overhead by 80 percent for startup, development, and configuration costs when onboarding new customers. It also projects a 20 percent reduction in ongoing running costs for this solution.

InfoCert is still implementing new capabilities on AWS, but it has already seen promising results. “New customers will soon start to use our new generative AI–powered solution in production. Simultaneously, we’re conducting a side-by-side comparison with our current system to facilitate a smooth transition for existing customers,” says Leoni. “We are quite confident that we will attain the same level of classification confidence in the near future.”