Genpact Drives Operational Efficiencies and Generative AI Adoption Using Amazon Bedrock and Amazon Titan

Genpact, a global professional services and solutions company headquartered in the United States, employs over 125,000 people across more than 30 countries. The company operates multiple locations worldwide to develop new solutions for improving customer services and internal operations.  

With the rapid evolution of generative artificial intelligence (generative AI), Genpact explored its potential to transform risk operations and internal processes. The company studied corporate functions such as human resources, finance, legal, and procurement to identify areas where generative AI could make a significant impact. The goal was to monetize successful use cases as customer-facing offerings. Habeebur Rahman, vice president of information technology at Genpact, explains, “We started identifying business challenges and ways generative AI could alleviate the issues.”

Around the same time, Amazon Web Services (AWS) introduced Amazon Bedrock, a fully managed service offering a broad set of capabilities for building generative AI applications. Genpact, which was already running multiple workloads on AWS and using Amazon SageMaker for machine learning modeling, immediately began looking into the new service. “Amazon Bedrock piqued our interest because, compared with other generative AI offerings on the market, it looked like the most powerful and cost-effective option,” states Rahman. 

Genpact registered itself as one of the early Amazon Bedrock customers. Genpact’s IT leadership team and colleagues attended events like AWS re:Invent, where Amazon Bedrock was discussed and use cases explored. “Although Amazon Bedrock was still evolving, it offered one of the largest token sizes among competing solutions, thus offering more accurate results,” says Munish Dargan, senior vice president of enterprise technology at Genpact.  

Like many experienced IT teams, the Genpact team was new to generative AI. Consequently, the developers selected by Genpact’s IT team to lead the project valued the support they received while working on Amazon Bedrock. Genpact worked with AWS Professional Services to expedite new project deployment and subscribes to AWS Enterprise Support for proactive assistance. This support optimizes infrastructure and costs, providing regular updates and technical training such as AWS Immersion Days and workshops. Dargan notes, “The Genpact IT team was happy to have access to AWS experts who provided guidance with sample code and were ready to answer any questions about generative AI. Overall, we’ve reduced the period spent on learning and experimentation by 10–15 percent thanks to the assistance from AWS.”

Genpact developed generative AI applications using Amazon Bedrock for two primary use cases. The first involved employee background check validation reports used during internal audits. Internal audit teams have to validate such reports for new employees or employees who have switched roles internally. Client requirements for background checks can include up to 20 criteria that may have been verified by separate vendors or across many individual files, involving searches through structured and unstructured data in multiple files. "We've reduced the time for background check validation reports by 50 percent—from 5 to 2.5 days—using our generative AI application built on Amazon Bedrock," states Rahman.

Furthermore, the standardized process saves time and reduces the likelihood of human biases or errors. Moreover, it enhances compliance and mitigates risk. “Although we still perform final checks manually, we've offloaded much of the time-consuming work that prolonged the entire operation,” Rahman explains.

For the second use case, Genpact developed Playground, a generative AI application for its 125,000-plus employees. For this application, Genpact deployed Amazon Bedrock alongside the Anthropic Claude 2, Stable Diffusion, and Amazon Titan large language models. The generative AI application processes text input and generates images based on user requirements.

Selecting AWS for generative AI supports one of Genpact’s primary strategic objectives. Indeed, the company ensures its AI-first solutions keep evolving so that users can deliver more quickly while improving quality. 

Using generative AI on AWS, the employee background check solution now provides comprehensive coverage, enabling 100 percent compliance checks. The machine accuracy for validation against client requirements has reached 80 percent, with ongoing improvements. Additionally, operational efficiency has increased by 75 percent, significantly boosting overall team productivity.

Beyond its accuracy, this solution stands out for reducing manual effort by about 90 percent. Automation has freed the Genpact team to concentrate on additional environmental checks and explore new areas like record-to-report. “We are transitioning into an AI-first organization, integrating this technology across all our solutions,” says Dargan. “We are incorporating the success of our Amazon Bedrock implementations into our client offerings to enhance our customers' performance.”

Using AWS, Genpact’s Playground is democratizing AI cost-effectively across the company’s workforce. Within nine months of its launch, Playground facilitated over 2 million interactions, helping employees complete their daily tasks faster and more accurately.

Comments Dargan, "Our future lies in making AI accessible to everyone in areas like data forecasting classification and anomaly detection so that we can deliver more value to our clients and outpace our competition." The company plans to continue working with AWS to develop new use cases for generative AI.