AI-first or nothing

If an operating model defines how an organization positions people, process, and technology to deliver customer value, then companies with an AI-first operating model are those that prioritize the use of AI to weave more intelligence and automation into the firm’s products, processes, and experiences.

Data gathered from 100+ global CIOs at the Metis Strategy Digital Symposium in July 2020 personifies the trend toward AI-first operating models: 66% of CIOs stated that they have teams focused on identifying AI use cases, conducting pilots and scaling those cases that improve outcomes. Of the CIOs who do not currently have resources focused on this, roughly 60% indicated it is on their roadmap. 

In our work with Fortune 500 companies, we have identified common characteristics among organizations that successfully navigate the shift to AI-first. Below are a series of smart first steps digital leaders can take to initiate, accelerate, or course correct their AI transformation.

1. Standardize the understanding of AI

Digital vanguards such as Mars Chief Digital Officer Sandeep Dadlani have taken strategic approaches to formalize the understanding of AI. In a July conversation with Metis Strategy, Dadlani shared that he deliberately delayed development of formal AI curriculum in favor of a business-centric approach that would “help people learn by solving problems in their context.” Teams approached every business and market and started by first understanding the problems that needed to be solved, and then applying concepts of AI and user centricity to solve them.

Rapid brainstorming sessions with senior leaders can also help deepen an organization’s understanding of AI. A simple matrix with AI technologies (e.g., supervised machine learning, unsupervised machine learning, natural language processing) on the vertical axis, and business problems on the horizonal axis can be used to frame the conversation. A facilitator then describes the art of the possible for each technology, and senior leaders brainstorm use cases to address existing business problems. Senior leaders can repeat this exercise within their own teams to ensure this understanding trickles down through the organization.

2. Define how AI drives busines outcomes and creates more ways to win

In its simplest form, an effective strategy answers two major questions: where to play and how to win. AI-first companies use AI to create more ways to win, such as delivering personalized experiences or creating capabilities that are exponentially scalable. As a digital leader, articulating the relationship between AI and business outcomes is critical to achieving buy-in.  Specifically, companies should define exactly how AI initiatives will impact all aspects of your business model, including customer value proposition (e.g., why customers buy), profit formula (e.g. how we achieve a profit), and key resources/processes (e.g., how we create and deliver value).

Knowing the answers to these questions is often a prerequisite for initial funding of AI initiatives, and yet many executives fail to explain in terms that resonate with business. We recommend a series of one-on-one meetings with key stakeholders to ensure a clear story about how AI investment can improve business models and drive targeted outcomes. Looking for inspiration on how AI can improve your business model? Check out our discussions with AI pioneers Sherif Mityas (TGI Fridays) and Vijay Sankaran (TD Ameritrade).

3. Identify cracks in your digital foundation that will limit your AI transformation

The promise of AI-driven business outcomes is alluring for senior leaders, but technical and organizational rigidities often stand between quick wins and true enterprise-wide transformation. Digital leaders must create a comprehensive roadmap that includes both investments to drive quick wins while concurrently strengthening the digital foundation. Here are a few recommendations to help implementation run smoothly:

Get rid of tribal operating models

Large enterprises, whether they admit or not, typically operate in “tribes,” defined by P&Ls, business units, and functions. Performance incentives are aligned with these tribes, driving an “us vs. them” mentality when it comes to strategic planning and capital allocation exercises. This results in a technology estate wrought with disparate data and applications, each designed to drive outcomes that benefit a certain tribe. AI-first organizations look beyond organizational silos and actively work to reconstruct the enterprise on a standard digital foundation. Microsoft, for example, rebranded IT as “Core Services” during their digital transformation to represent its new longitudinal, BU-agnostic mission to provide the components that the organization can use to build processes that run the whole company.

Centralize and standardize your data to get a holistic view of customers

The disparate (and often incompatible) applications and data sources that arise across organizations as a result of tribal structures make it difficult to connect the dots between different customer interactions. Many organizations have acknowledged this shortcoming and have launched multi-year programs to build an enterprise-wide data platform (or “data lake”) to ingest, contextualize, and integrate all sources of data and develop a 360-degree view of the customer. With a trusted data source that spans the organizational processes, a company can develop an AI application using a supervised learning algorithm to preemptively identify customers that are likely to churn and proactively drive outreach to at-risk customers. 

Focus on re-usability rather than single-use integrations

Organizations have traditionally built integrations and functionality for specific business units or functions, often leading to redundant functionality, tightly coupled custom integrations and compounding complexity on the technical estate, ultimately increasing total cost of ownership and slowing speed to market. AI-first organizations harness the power of reusability and modularity to create a centralized layer of application programming interfaces (APIs) that expose data and libraries of software components to enable rapid development of AI applications across the enterprise. In this model, application development becomes less about building new functionality from scratch, and more about orchestrating existing, complementary functionality to drive scale and scope economies in the business.   

Solving for these rigidities is both costly and time consuming, but ignoring them undermines their potential to enable transformation. Take the time to assess these dimensions of your organization and devise a long-term roadmap to remediate cracks in the digital foundation.

4. Dedicate resources to identity AI use cases and conduct pilots

While shoring up your digital foundation, dedicate teams to focus on identifying AI use cases that drive impact and organizational learning. Allstate, for example, stood up an AI center of excellence in which cross-functional teams use a combination of value stream mapping and design thinking to identify AI opportunities and define AI-driven solutions. “The cross-pollination between stakeholders is great as the AI architects can demystify the process and become more knowledgeable about the business, while the process SMEs [subject matter experts] become more informed on the use cases for AI,” CTO Chris Gates said at the Metis Strategy Digital Symposium.

In identifying AI use cases, we recommend teams focus on scenarios in which AI can make products, processes, or experiences more intelligent, more efficient, or more scalable. A healthcare system, for example, may deploy an AI solution such as CurieAI to inject intelligence into the process of monitoring COVID patients’ breathing patterns to proactively identify early indicators of deteriorating conditions. To drive efficiency, the same hospital may deploy AI-enabled chat bots through a solution such as LifeLink to help patients navigate the waiting room experience. 

The ability to embellish products, processes, and experiences via AI is particularly exciting because the opportunities are limitless. However, it is important to not overlook the customer implications of automation decisions. In a healthcare setting, even when patients are physically surrounded by technology at their bedsides, “human interaction is absolutely fundamental” to the experience, says Paola Arbour, CIO at Tenet Healthcare. She added that there are certain elements of business models that should be preserved, in this case the patient -physician interaction.

Paola and her team look to AI to enhance the way physicians interact with patients, by providing them with the right contextualized data at the right time to drive the right outcome. This formula, often referred to as “Human + Machine,” is applicable in many scenarios where human interaction is critical to delivering optimal experiences. 

As your list of opportunities grows, we recommend conducting pilots for those with the most potential to drive targeted business outcomes and the highest degree of feasibility.  This scoring exercise should first be done rapidly by teams focused on AI, then validated by a digital steering committee (or similar designated body) to obtain “seed funding” and drive MVP (minimum viable product) development. MVP results can then be assessed by the digital steering committee to determine if measurable value was captured and if there is justification for scaling the initiative.   

5. Scale MVPs that improve outcomes and begin continuous learning

Those MVPs that capture value and show the potential to scale should ramp up to formalized delivery programs. As MVPs are scaled to more functions and customer segments, more data will come into the pipeline, which means organizations have more levers to refine their algorithms. Better algorithms, by extension, can lead to better experiences, and ultimately more usage (and then the virtuous cycle starts all over again).

Consider a subscription grocery delivery app that has recently been enhanced with personalization features. The intended outcome is to increase average basket size. As users add things to their basket, the algorithm identifies items that customers with similar baskets have ordered and displays them on the screen as suggestions. Customers can then accept or reject the recommendations. As the sample size gets larger and the algorithm analyzes additional customer data (e.g. demographic, order history, behavior, etc.), the algorithm gets better and better at converting recommendations to sales. This in turn leads to better shopping experiences for customers, more usage of the app, and more data to further refine the algorithm.

An inflection point for digital leaders

Artificial intelligence is undoubtedly changing the business landscape and creating new ways to win for digital native and digital immigrant businesses of all sizes. Digital leaders must ask themselves if they are positioning their businesses for success in the new data-driven world, in the short term and the long term, across people, process, and technology. Technology visionary Tom Siebel, founder of Siebel System and CEO of C3.ai, warns that the century ahead will be one of “corporate mass extinction” and those that fail to acknowledge the shifting landscape may end up like the 52% of the Fortune 500 that have fallen off the list since 2000. So what will it be:  AI-first or nothing?