AI in Business: Act Quickly to Maximize ROI

The Urgency of Acting with AI for Quick Return on Investment

In today's business world, the adoption of artificial intelligence (AI) can no longer afford to be a slow and theoretical process. To leverage AI effectively, it is crucial to focus on concrete use cases and proceed through trial and error. This approach not only generates a quick return on investment (ROI) but also fosters team buy-in.

Too often, companies hesitate to embark on AI projects without having clearly defined their needs. However, this approach can prove counterproductive. Writing complex specifications based on assumptions or unrealistic expectations regarding AI can distance companies from the reality of their daily operations and current technical possibilities. In contrast, embarking on a concrete project, even a modest one, can position a company advantageously in the race for competitiveness.

The Challenges of Internal Process Understanding

A major obstacle to AI adoption lies in the lack of understanding of internal processes by management and leadership. In many companies, particularly in the service sector, this lack of knowledge can hinder the identification of automation and optimization opportunities. For example, a service company discovered that an external provider was charging millions of euros to resolve ambiguities in billing, ambiguities caused by poor management of the supplier database. A thorough analysis allowed the company to eliminate this costly task by streamlining the database.

The Importance of Iterations in Developing AI Agents

In another case, a service company sought to optimize the deployment of its employees across multiple sites. The development of an AI agent for scheduling and dispatch required several practical iterations. These iterations allowed the generated schedules to be compared with management's expectations, thereby revealing implicit business rules. Through this process, the company was able to refine its operational constraints and adjust parameters, such as actual training costs, to improve service quality.

The Difficulty of Imagining the Possibilities Offered by AI

In addition to the difficulty of precisely defining its needs, it is even more complex to imagine what AI can actually accomplish. The available technologies are numerous and evolve rapidly. Generative AI, well-known through chatbots, is just the visible part of a broader set of technologies. Companies can utilize multimodal models, machine learning, and autonomous reasoning technologies to automate tedious tasks or assist employees in complex decision-making.

Testing Technologies to Validate Their Effectiveness

To fully leverage these technologies, it is essential to test them in real-world situations. This allows for the evaluation of the validity of the results obtained and understanding the conditions necessary for their success, including speed and associated costs. By defining a clear scope of action, such as a persona or specific workflow, and acting quickly, companies can achieve tangible results that build confidence and facilitate the broader adoption of AI.

Towards an Agile Approach to AI

Rather than getting bogged down in writing theoretical specifications, companies should adopt a more agile approach. By focusing on concrete use cases and adapting to available technologies, they can not only achieve quick returns on investment but also lay the groundwork for a broader and more effective deployment of AI in their future operations.