Executive’s guide to developing AI at scale

Developing artificial intelligence and analytics applications typically involves different processes, technology, and talent than those for traditional software solutions. Executives who possess a solid understanding of the basics can ensure they’re making the right investments in their tech stacks and teams to build reliable solutions at scale. We’ve created an interactive guide to help.

October 2020, Interactive Guide

Defining environments

Lab

Due to its experimental nature, analytics development work—including data exploration, experimentation with predictive models, and development of prototypes through rapid iterations— must be performed in a “lab” environment that’s separate from other systems so that it doesn’t hinder normal business operations. Lab technologies must be flexible and scalable to handle the changing demands of the analytical approach (eg, new data, new modeling techniques) and modular to enable developed solutions to port to the factory through DevOps.

Factory

After development in the lab, analytics models move into the “factory” which provides an environment for running analytics jobs 24 hours a day, 7 days a week, 52 weeks a year. In order to put a solution into production at scale (ie, making it regularly and reliably accessible to users), it has to be robust (able to handle typical errors, including variances in incoming real-world data), maintainable, executed efficiently through continuous deployment processes, and integrated with core systems, and it must include performance management and risk controls to avoid any detrimental impact on operations.

Ways of working

MLOps

MLOps refers to DevOps as applied to machine learning and artificial intelligence.

Short for “software development” and “IT operations,” DevOps is the application of software engineering practices to IT operations, such as packaging and deploying production software.

MLOps aims to shorten the analytics development life cycle and increase model stability by automating repeatable steps in the workflows of software practitioners (including data engineers and data scientists). While MLOps practices vary significantly they typically involve automating integration (the frequent checking in and testing of code) and deployment (packaging code and using it in a production setting).

Roles

The work in the lab and factory is carried out by cross-functional teams made up of data and software professionals (eg data scientists, machine learning engineers, cloud architects) as well as business professionals with varying levels of data science expertise (eg, subject matter experts, translators).

Explore processes & technologies

Start the exploration

About the author(s)

Nayur Khan is a senior expert in McKinsey's London office, Brian McCarthy is a partner in the Atlanta office, and Adi Pradhan is a consultant in the Montreal office.


The authors wish to thank Mayur Chougule, Joe Christman, David DeLallo, and Maxime Delvaux for their contributions to this content.

We strive to provide individuals with disabilities equal access to our website. If you would like information about this content we will be happy to work with you. Please email us at: McKinsey_Website_Accessibility@mckinsey.com