Genpact Unveils Industry-First Operating Playbook for Scaling Responsible Generative AI Projects

Sreekanth Menon, Global AI/ML Services Leader, Genpact
Sreekanth Menon, Global AI/ML Services Leader, Genpact

(US and Canada) Genpact, a global professional services firm focused on delivering outcomes that transform businesses, has unveiled an industry-first playbook:  FMOps – The Generative AI Imperative for Production. The playbook, which focuses on Foundation Model Operations (FMOps), provides practical guidelines for building solid and efficient foundations for generative AI solutions.

"As Generative AI reshapes enterprises across sectors, we continue to find that clients need clear directions for building ethical and scalable generative AI solutions," said Sreekanth Menon, Global AI/ML Services Leader, Genpact. "This playbook offers that blueprint, guiding the transition from pilot generative AI projects to production."

Developed in collaboration with Nasscom, a non-profit industry association and the apex body for India's $245 billion technology industry, the playbook delves into the fundamentals of FMOps. It meticulously outlines their operational benefits within AI systems:  improved collaboration, faster release cycles, enhanced efficiency, rapid model deployment, seamless scalability, and risk reduction. The playbook also provides detailed discussions on FMOps, Large Language Model Operations (LLMOps) and Machine Learning Operations (MLOps), including:

  • Understanding the distinctions of generative AI from traditional AI

  • Recognizing the significance and importance of FMOps

  • The role of LLMOps within FMOps and generative AI

  • Key differences between MLOps and LLMOps

The playbook takes an in-depth look at LLMOps, which serve as the foundation for operational capabilities and infrastructure required to deploy generative AI solutions, and the steps necessary for successful implementation:

  • Identifying the right model, technique, team, and technology stack

  • Understanding the current LLMOps landscape

  • Creating a Responsible, Accountable, Consulted, and Informed (RACI) framework for LLMOps implementation

  • Operationalizing Large Language Models (LLMs) with LLMOps

  • Selecting appropriate metrics

  • Setting up policy management guardrails

The collaborative effort between Genpact and Nasscom builds upon previous work on MLOps best practices, a set of principles that help technology teams develop, deploy, monitor, and scale AI models effectively.

To download a complimentary copy of the playbook, visit: Moving Generative AI from Possibilities to Production.

Related Stories

No stories found.
CDO Magazine