Managing AI projects is a complex task due to the highly sophisticated technology involved. According to a study by Rackspace, only 17% of companies, or every sixth company, are able to advance an AI/ML project to a sufficient level of maturity. The success rate is even lower than that of simple data projects such as creating data warehouses. Therefore, it is essential to approach official press releases about the introduction of AI with caution. At best, the company may raise the level of maturity of its ML features in its product. Experienced managers understand the need for multiple attempts to achieve real success.
Highlight the conflict:
- the engineering team does not have the necessary business skills
- the management team does not have the necessary technical skills
I would like to highlight an intriguing and intricate conflict/contradiction:
- the engineering team does not have the necessary business skills
- the management team does not have the necessary technical skills
The difference in understanding between the two groups regarding expected outcomes, cost, and timeline poses a significant risk. This risk may lead to loss of money, time, and even the team itself.
In this scenario, as practiced at 1Dataline, three vertical functions can be carried out simultaneously:
- Data collection
- AI model development
- Operational use the brand new AI/ML product in real environment
Vertical teams have the ability to manage the life cycle of their sub-product within their respective segment, allowing for more independence. The culmination of these sub-products creates a functional workflow.
Here is a successful recipe. Essentially, we need:
Step 1. We need to establish a new corporate management and organizational structure within this Department. The new roles, procedures, and responsibilities should be redesigned. To achieve this, we should first sit down and describe the necessary details/steps. Afterwards, we will need to recruit the appropriate personnel. As this typically requires a new search in the HR market, we should avoid retraining existing staff.
Step 2. Please ensure proper training is provided, as each of the three verticals requires a pair:
- Manager with a solid grasp of technology.
- IT specialist who has a good understanding of the unit economics of the product (should have a grasp of the business).
Step3. Establish a well-functioning operating process that includes understanding who and when to train, when and how to collect data, and when to test and put into operation. It is crucial to have integral and systematic planning and to work with dependencies.
