Setting Up Your AI Function - Lessons Learned
About a year ago, I was asked by a long time friend and business partner Mark Minevich (a renowned visionary in the field of AI) to work with a company that was looking to better understand the strategic benefits of AI, where it stands today, what it means to its customers and its own businesses. Is AI at a level of maturity to take it seriously? I was introduced to Alexis Lope Bello, a progressive but careful CEO of a company called Comtrade that provides leading IT products and services to a variety of industries. Coupling Mark’s prolific involvement in AI, and my strategic, corporate development experience in establishing new capabilities in emerging trends, Alexis gave us the go ahead to investigate. So now, twelve months later, Comtrade has a fully operational division focused on leveraging AI's best practices and innovations for itself and the customers that Comtrade serves. Getting to this point however, was no walk in the park.
One of the first challenges was to sift through the hype that was going around from sell-side companies labeling everything AI. This bastardization created a blur of understanding of what AI can really bring to the table. The outcome has been a number of false starts and a misrepresentation of potential outcomes for those brave companies that completely drank the Kool-Aid. Avoiding the hype train was the first thing to do. While strong consultancies and leading universities have generated useful thought leadership materials around the strategic nature of AI and its core disciplines, not much has been written about how to properly implement AI strategy from an organizational, process and technical perspective.
When focusing on the ramifications of implementing AI, key issues emerged. As an example:
IT organizations and the businesses they support must alter its interaction model due to the increased involvement needed by the business to train AI. The level of involvement reminds me of the days when 15 years ago, the quantitative analytics trading departments often got reshuffled between the IT and business organization until it eventually saw its place better served as a decentralized group within the business. The implications for IT were profound and humbling; but both parties made to work as process and lines of responsibility /accountability had to be clearly drawn. Depending on the industry you are in, the AI responsibility may need to be separated between the business and IT. I’m sure I will get an earful from my CIO colleagues who read this
IT hasn’t done a good job at explaining to its business partners what AI really is about and what it means to their own organization. I recently had a healthy discussion about AI with an executive committee member from a global Fortune 50 corporation. He had received debriefs from top consulting firms describing the benefits of AI; it being a boardroom imperative that should be a top priority. But what wasn’t discussed by these top consultancies were the challenges associated with AI deployment, its transformative effects and the responsibilities the business units had to its success. It’s much more of a business issue than most realize
The key to a successful program is not to hoard or to hire only data scientists for your AI program. Without the proper methodology, and understanding of new people roles that will be required to manage data set acquisition, new methods of testing and new development environments, your data scientist team will get underutilized or saddled with work they are not qualified for, resulting in a higher churn rate with subpar outcomes
Ready access to qualified researchers and data scientists are far more inflated by sell-side vendors. One vendor that I spoke to said that their firm cornered the market in having over 1000 data scientists available for analytics and AI in India. This is a doubtful assertion. In the United States, there are only an estimated 5000-6000 qualified data scientists and the high tech companies (Google, Microsoft, Amazon, IBM, etc..) have most of them. Managing scientists also introduces new cultural challenges that are not fully understood by commercial IT and HR organizations
To solve these issues, we installed an extremely cohesive partnership network that goes beyond name and is an operational extension of Comtrades overall delivery model. Comtrades AI division is a separate centralized utility (apart from IT) and all businesses across the company leverage it’s resources, IP and capabilities. There is a deep synergy and interdependence between the two with complimentary responsibilities that are shared. A governance structure is in place where partners are encouraged to contribute to the strategic direction of Comtrades AI strategy. The AI Advisory board (chaired by yours truly) ensures that it’s partners are involved in client meetings, deal opportunity rationalization and other various types of business development activities.
If you are in a position to influence, approve, lead the development of AI for your organization, it is no doubt an exciting time. Like all new paradigms, integrating AI into your organization is full of potential pitfalls, inefficiencies and ROI questions. As many have written, the payoff for a successful AI capability is great. Until steady state is reached however, the transformative impact of AI at full velocity will be a rocky road.
Building out a sustainable capability is not something that you can just slap together. Having said that, examples of firms leveraging AI are emerging every week, too many to count. These examples range from small to big. Embedded in your google searches will be false claims of products and services using AI; a claim that will be nothing more than the words inside of a sales brochure. Buying an AI solution might seem like a short cut and prudent approach but, it could also mean being baited with vaporware that will set you back months at a time.
Albert R. Eng - Global Head of AI - Comtrade
Albert also acts as the CEO of his own firm - Corporate Development & Strategic Advisors
