5 Essential Steps for Successful AI Projects

5 Essential Steps for Successful AI Projects

Delivering lasting business impact using AI? Based on experience from AI projects across the globe, Huw Kwon, Global Head of Artificial Intelligence at Cognizant, outlines five important steps to industrialize AI.

Recently, key players from the financial industry gathered at the Findig event in Norway. Huw Kwon was one of the speakers with a session about ”AI, from talking, walking to running” focusing on how to deliver real business benefits using AI. 

Download eBook: AI in Nordic enterprises

He experiences that executives, independently of industry, are facing the same sort of challenges connected to AI projects. It’s about difficulties scaling up, getting any output from data modernization initiatives such as data lakes, and creating business value from AI use cases. 

Typical failures include a lack of holistic approach; only tech people are involved instead of tech and business representatives, lack of understanding of AI’s impact; the belief on its potential is too exaggerated, competence issues, and a too strong focus on data only. 

How do you make AI work like a pro then? Well, according to Huw Kwon there are five essential steps to make business impact that no successful AI project can skip:

  • Identify the source of value. To accomplish this, you have to have qualified people like data engineers and data scientists. Talk it through thoroughly to fully describe the source of value, to identify performance goals, and build hypotheses in data and analytics.
  • Data modeling. Dive into the data ecosystem; collect relevant data, build data models, and validate your data. Truth is, that most data isn’t ready to be utilized. 
  • Classification modeling. This is where the clever stuff is happening; model building, finetuning of performance, and estimating of business model.
  • Impact delivery. This is really important – here you apply your model to business performance. At this stage, some experience they have used the whole budget already. Make sure to measure cause and effect. Refine your models if needed.
  • Industrialization of AI project. At this stage, you ripe the benefits of the previous steps. You productionalize your AI project, focus on change management and automation. 


How can you make all this happen? It might be hard to accomplish it all within your company walls. Cognizant offers a structured approach in so-called PODs, consisting of skilled, cross-functional teams with a strong focus on business outcomes. All phases in the project are measured to ensure progress and to uttermost make your AI project run.

To get a more in-depth view of these steps, watch this video with Huw Kwon’s session and have a look at our AI section of the web.

Related Publication

What Does it Take to Lead in Digital Times?

Digitalization, upstart competitors, the need for breakneck speed and agility, and an increasingly diverse and demanding workforce require more from leaders than what most can offer. Check out our new survey with 4,394 global leaders.