top of page

Bridging the Gap During AI Adaptation

Bridging the Gap During AI Adaptation

Although artificial intelligence has gained popularity in the past few years, not every company is ready and able to implement AI in its products, services, or processes. The reason is the massive disconnect between the business domain, the engineering domain, and the AI science domain.

I see that in the upcoming years, AI leaders such as AI product managers with AI strategy know-how will play a pivotal role in closing this gap.

In this article, I'll explain why there is a disconnect between these domains and what techniques you can use to ensure alignment across them. If you want to learn more about leading AI products and AI teams, join me in one of the AI Product Institute courses at

Let's start with the business domain. The business domain doesn't have much in common with the other domains; a business is an activity of earning money by producing or buying and selling products. Business concerns range from profitability, sustainability, brand image, cost structure, customers, to compliance. Even though a company consists of many activities originating from production, partnerships, customer relations, and support, profit is the key metric used to determine the value created by a company. 

On the other hand, the AI science domain deals with fulfilling a condition through learning stated by the problem. This domain's concerns are usually about scientific techniques and algorithms such as natural language processing, computer vision, reinforcement learning, deep learning, and machine translation.

Somewhere in between, we have the engineering domain. Similar to the other two, the engineering domain also has minimal overlap. This domain is primarily concerned with solving problems such as scalability, security, reliability, performance, capacity, and ETL through the use of software and hardware technologies.

The real challenge that organizations face is closing these gaps by connecting the business domain, engineering domain, and the AI science domain for sustainable profitability. On the one hand, we have teams from business, marketing, sales, and logistics, whose primary objective involves satisfying revenue, cost, and customer-based goals. On the other hand, we have engineers and scientists whose primary aim is to develop technical solutions and research better algorithms. These three different disciplines may speak the same language, but they don't share the same goals. Therefore, most companies struggle with utilizing AI for sustainable profitability. 

The problem is two-fold: first, many teams ignore the difference between short-term ventures and long-term ventures. Second, there is no shared KPI between business, engineering, and science.

The first problem originates from trying to enter the AI era by placing all of the efforts into science without thinking how to engineer these solutions, nor how to make a sizable impact on the business. This approach is what we call AI for AI's sake, and the philosophy did not change since the PC era. Many organizations with a lack of AI leadership overlook this point and start their AI transformation from the technology side. This approach is a bottom-up where the engineer or data scientist decides the future of the organization.

To overcome this problem, organizations need to define corporate-level strategies to plan and coordinate long-term AI ventures, mid-term emerging AI opportunities, and short-term core business AI operations. In other words, they need to understand the difference between the core business, emerging business, and future business. And invest in long-term AI ventures as much as they invest in adapting emerging AI solutions to core business.

One of the tools to make the correct investment at the right time and continuously bring innovation to the customers is the McKinsey's three horizons of growth framework.

Today, companies like Amazon, Google, Microsoft, and Apple have succeeded in building these three horizons of growth and are integrating future technologies one after another to provide unprecedented customer value. If you want to learn more, please check out the framework at

Based on my experience, three horizons of growth framework can be applied at every level of the product and company. For example, consider these three AI projects in a large organization: 1) AI-based production quality assurance, 2) customer's next-best-action prediction for the marketing department, and 3) AI-based contract review for legal departments. A brief discussion with the data science, engineering, and business teams would quickly reveal that only the first AI solution provides a sizable business impact in the short-term and the best cost-benefit. While #2 and #3 sound "cool," they require considerably more investment in the long-term. Therefore, only the first initiative is a candidate to be utilized in the core business. Of course, we are assuming here that the organization possesses the capability.

On the contrary, an investment in the other two initiatives depends heavily on corporate strategy, which in turn depends on the company's right to grow.

An effective way to solve the second problem - the absence of crosscutting KPIs - is by employing the objectives and key results (OKR) framework not just at the department level but also at each horizon level. It is at each horizon level, because the core business objectives tend to be utterly different from future and emerging initiative objectives; while core business strives to run the current business like a well-oiled machine and thereby to target profitability, the future initiative aims experimentation and discovery to find a spark to light a business and thus operates like a startup.

The first aspect of OKR is objective: organizations need to define a shared objective across the business, science, and engineering. This will align the whole organization in the same direction, in the same way as nearby atoms align along the same north-south field lines in a magnet. The second aspect of OKR is key results: organizations need to define a common language across all units for measuring the progress. Take, for example, an online credit card fraud detection solution of an e-commerce company. Below you can see the topics the objectives of the core and future business will consider. At the cure business organization level, for instance, there will be four or five objectives such as profitability, customers, employees, research. On the contrary, at the future business organization level, there will be objectives to find new viable business ideas.

Although adapting the three horizons of growth and OKR frameworks is not an easy task, it is crucial for finding AI opportunities. These frameworks allow allocating the right resources at the right time to the right project. Research, for instance, is a costly investment that needs to be part of future investments without short-term expectations. Of course, if the company has enough resources to fund a hefty research bill. Even though it sounds self-evident, almost every company I've consulted in the past wanted me to develop their futuristic AI idea to improve their current business. This is primarily due to the lack of end-to-end visibility.

Most roles typically have isolated responsibilities in organizations. However, defining the right AI investment strategy requires having end-to-end visibility across all teams and initiatives. Therefore, I see that only the AI product manager is in the position to bridge the business, science, and engineering over core business, future, and emerging initiatives.

If you want to learn more about leading AI products and AI teams, join me in one of the AI Product Institute courses at

45 views0 comments

Recent Posts

See All


bottom of page