So you want to start using AI at your company. Now what? 

First, evaluate if it has an appropriate place in your company. Many organizations hire a data scientist or an entire AI team with an anticipation of a fast, massive, magical gain. Even though by now most people realize that these expectations are naive, the general public and even venture capitalists are still attracted to the idea of miraculously making everything better with AI. After all, it is tempting. 

When deciding whether or not to start using AI at your company, realistically consider how much real value it might bring. There are two questions that you should be asking yourself. First, what problems will it help me solve? Second, do I have or can I obtain large quantities of clean parsable data to enable it? 

In order to move forward, you need to have a clear answer to the first question and a positive answer to the second question. Consult with an expert and formulate your use cases. Consider the data that you have, or might start collecting, and the level of expertise and bandwidth of your existing employees. Some straightforward and small-scale AI systems are easy to build with automatic machine learning (ML) tools, provided the problem statement is clear and you have relevant, abundant, and clean data. 

These off-the-shelf systems can help generate the momentum needed to prove that AI can bring value to the company and convince stakeholders that investing in it is a prudent decision. You will still likely need someone who is well-versed in machine learning and data, but they do not have to be an AI guru, and you definitely do not need an entire team. Most of the effort in the case of a small-scale project is typically focused on generating, cleaning, and maintaining the datasets that the AI is learning from. For a larger and more complex AI system, you would need to grow your team. A common pitfall is to keep hiring data scientists. After all, they may have proven the initial value of your AI, but at the growing stage, you’ll need to invest in other roles as well: data engineering, data infrastructure, and, potentially, an in-house ML engineer. If you hire data scientists without adequate engineering support, you will be left with many concepts that never become products. 

Another important component of building and productionizing successful AI at a larger scale is leadership buy-in. Without support from the top, projects will get stifled. After an AI project has been prototyped, there is still a long road ahead towards a production implementation. This will require contributions from engineering, product, design, QA, and other teams. If leadership is too focused on the current operations and short-term gains, and not on the long-term benefits of the automation and predictive powers that AI provides, no large-scale project can be implemented. 

Bringing AI into your company is no easy feat and can easily lead to wasted time and effort. But with a clear objective, abundant and clean data, and a mindfully built team with leadership support, AI can transform your company.