ºÚÁÏÃÅ

Ìý

Introduction: Too fast, too soon?

Ìý

In their urgency to get ahead or even just keep up, some organizations are launching AI initiatives on shaky ground. More than next-generation technology, AI needs to be supported by a clear vision and a robust platform that provides ready access to carefully vetted, highly usable data to create products that generate business value.ÌýÌýÌý

Ìý

In this issue of Perspectives, ºÚÁÏÃÅ experts and one of the leading figures behind BMW Group’s industry-leading Connected Vehicle AI Platform share their views and recommendations on building a scalable, cost-efficient and future-ready foundation for applied AI that achieves great things, for both internal users and the organization’s customers.

Ìý

Significant proportion of tech and business leaders are skeptical of their organization’s data quality

Source: Salesforce

Ìý

Ìý

Ìý

i. Putting data assets to use

Ìý

Delivering tangible business outcomes has to be the primary objective of any AI initiative, and that requires organizations to prioritize use cases that represent meaningful improvements for the customer base, before creating an AI-ready data foundation that makes those enhancements possible.Ìý

Ìý

At BMW Group, providing proactive customer care was the principle guiding the design of the company’s AI platform, which now enables teams across the company to develop, maintain and operate AI use cases, and reduce time to market for AI applications that enhance the way customers interact with and feel about their vehicles – and the BMW Group brand.

Ìý

Ìý

ii. Tech elements of an optimal foundationÌý Ìý

Ìý

While the specific setup of an AI or machine learning-ready platform may vary across organizations, effective platforms share several core characteristics – chief of which is the ability to access and make use of consistently high-quality data.Ìý

Ìý

Once a foundation of quality data is in place, usability becomes a key determinant of the platform’s success over the long term, as even the best teams will face challenges with training and maintaining models, and bringing innovations to production.

Ìý

Components of an AI/ML-ready data foundation

Source: ºÚÁÏÃÅ

Ìý

Ìý

Ìý

Ìý

iii. Scaling up while controlling costs and maintaining standards

Ìý

As organizations begin leveraging their custom AI-ready data foundations in earnest, scaling up can still be a difficult, and expensive, endeavor.Ìý

Ìý

As BMW Group’s experience shows, teams can manage the process by constructing a repeatable setup that can serve multiple use cases to reap economies of scale; conducting a level-headed assessment of the costs and tradeoffs involved in managed versus internally delivered services; and putting guardrails in place to govern data privacy, security and performance. Automation is the key to making this possible and a decisive factor in a platform’s ongoing viability.Ìý Ìý

Ìý

iv. Cultivating buy-in, ongoing education: Platforms as a processÌýÌý

Ìý

While an AI-ready data foundation may boast innovative features and attractive potential benefits, the true test of its value lies in the extent to which it is adopted and utilized to develop products that are in turn embraced by customers.Ìý

Ìý

Organizations can pursue a multi-track approach to cultivating adoption and buy-in, from incentivizing ownership of data products, to putting in place comprehensive onboarding programs that address the requirements of different user groups. Ultimately ongoing engagement is best served by a human touch, whether in the form of firm executive sponsorship or forums for dialogue to address concerns fueling internal resistance.