Fractal Analytics Reduces Call Handling Time by Up to 15% with Generative AI on AWS

Fractal Analytics believes human imagination is at the heart of every decision, whether or not that decision is aided by artificial intelligence (AI). The global firm, one of India’s original AI unicorns, uses technologies including generative AI to empower imagination with data-driven intelligence.

Fractal offers an array of customizable solutions to improve customer and business outcomes across industries such as finance and healthcare. Patient Jarvis, for example, is an AI-powered data accelerator that puts a dynamic view of patient healthcare history at practitioners’ fingertips.

One of the company’s latest innovations, Knowledge Assist, utilizes large language models (LLMs) to make knowledge retrieval more efficient within large enterprises. Traditionally, data retrieval and consolidation from numerous internal sources has been a highly time-consuming task for knowledge workers. This data is often unstructured, increasing the complexity of each query.

Fractal’s client base comprises numerous Fortune 500 companies, so data security and privacy are paramount when developing new products. Most clients deploy Fractal’s solutions within their private networks, so platform flexibility is also important. Having already executed many large-scale software as a service (SaaS) implementations on Amazon Web Services (AWS), including a chat solution that handles 15 million queries per month, Fractal chose to build Knowledge Assist on AWS.

The company deployed its LLMs on Amazon Bedrock, a fully managed service with a broad set of capabilities to build generative AI applications with security, privacy, and responsible AI. Fractal’s engineering team experimented extensively with available models, evaluating cost and performance against their intended use cases.

Ritesh Radhakrishnan, client partner for Products and Accelerators at Fractal Analytics, says, “The generative AI space is evolving rapidly. Being able to choose from various LLMs on Amazon Bedrock, which we can swiftly implement or experiment with, along with the ability to use the platform as an API without hosting concerns, helps us experiment and scale faster.”

Fractal worked closely with AWS during the development process, receiving valuable guidance including benchmarking service costs. “For nearly seven years we’ve gotten constant support from AWS, whether for getting credits for experimentation, conducting technology workshops and training, or simply having access to experts who can help with a deeper understanding of AWS services and AI best practices,” adds Radhakrishnan.

In addition to Amazon Bedrock, Fractal used Amazon Elastic Container Service (Amazon ECS) to build connectors for Knowledge Assist and Amazon OpenSearch Service for vector/semantic search. The SaaS application layer runs on Amazon Elastic Kubernetes Service (Amazon EKS) and AWS Lambda serverless compute. Radhakrishnan comments, “The robustness of each of these services helps us focus on the application, rather than having to worry about the scaling and stability of the underlying framework. That speeds up our entire development process.”

Knowledge Assist upholds the same stringent security and privacy requirements that make Fractal trusted by its clients. The platform uses private endpoints to protect data within each client’s network, and data is encrypted end to end. Additionally, personally identifiable information is masked before storage in the application’s analytics layer.