Data science and AI

Trend 9: AI and ML find significant usage in cloud services

Analyzing massive amounts of unstructured data like images, transcripts, and recorded speeches is a top requirement for enterprises. It necessitates significant upfront investments in the compute and storage infrastructure to collect, cleanse, and tag the data for training and model building. Public cloud service providers (CSPs) saw an opportunity to provide pretrained models related to vision, speech, and language as platform as a service (PaaS) models. CSPs now offer fully managed cognitive services like AWS Comprehend Medical, Azure Form Recognizer, and Google Video AI. These services allow businesses to build cost-effective and faster-to-market solutions based on the latest AI/ML pretrained models.

Other cloud services like storage and information as a service (IaaS) virtual machines will complement the solution build phase. But PaaS services imply a lock-in with CSPs. Proper design and use of containerization and cloud-agnostic code-building platforms create an easier exit path, if required. Developers and software service providers are building mature domain-specific solutions using CSP PaaS services and making them available in the analytics and AI marketplace.

Infosys uses the Azure Form Recognizer for automated processing of prior authorization requests received through fax or email in Infosys Healthcare Platforms.

Data science and AI

Trend 10: Enterprise-level AI shifts from fragmented to integrated and managed activities

Platforms and services that perform all the functions of a typical AI/ML implementation lifecycle help businesses move toward a more standard, managed, integrated and collaborative environment. Platforms like H2O Driverless AI, Azure ML, and Amazon SageMaker bring citizen data scientists and CSPs together. They collaborate on everything from ideas and code to implementation and best practices. The maturity in the platforms and services reduces concerns around people experience, complex use cases, and shadow IT scenarios. The auto ML feature of the platform reduces the effort in identifying the right algorithm for use cases. Users utilize the Integrated Ensemble feature to produce one optimal predictive model for a use case. IT service providers like Infosys built their own accelerators and tools like data advisory and AI workbench to aid clients' AI/ML journeys.

A leading US-based bank, in association with Infosys, built a self-service analytics tool to generate daily actionable insights of complex, multivariate data for customer segmentation, profitability, and campaign analytics to enhance its mortgage business.