For every stakeholder, AI should feel simple. If it does not feel simple, then there are likely problems somewhere in the delivery. This could be issues with communication, challenges to implementation, the wrong infrastructure, poor project management, or the very strategy itself.
AI is an inherently challenging technology deployed to perform all manner of tasks. However, it should not matter whether the task is basic automation or deep learning, the importance of getting everything right should push your project to have the same attention to detail as a space launch. This is not to say you should overcommit resources. The key is that every team member understands both the details of their job alongside the anatomy of the project itself.
The first question is always what are you trying to achieve? The answer will require collective agreement across the organization, or at the very least within the team tasked with overseeing the project. It is also critical to ensure the proposed automation is the right answer, and part of a broader strategy that works to ensure any AI project fits within a clear series of parameters that account for employee skills, system integrations, department workflow, and other major projects planned or currently at work.
However, the most important starting point should be an honest discussion about whether the process is worth automating, or whether the technology required is deliverable as a whole product. This is critical for two reasons. First, if a solution requires additional software development or systems integration, then it may prove an unnecessary pain in the backside. The last thing you need is a substandard implementation because the product fundamentally was not ready for the task at hand. Second, you need to be clear that the process of automation is the right one. Automating bad processes makes life easier, but it is very hard to change that process once automation has occurred. This is not a game of chance you want to play.
It always bears repeating that it is important to ensure the project has strong support from leadership. Once there is board and c-suite support, clarity is needed to confirm that appropriate resources will be committed to the task. Some firms are fortunate to have available employees with the time and patience to take on the project management and development roles required to take the project from strategy to implementation with minimal problems. However, this is very rare and its worth planning for assistance from a services provider that has experience working with AI technology, the data in question, and your department function or industry.
Finding a Solution
More integration, less bells & whistles. Keep this in the back of your mind when you next find yourself in being pitched by an AI vendor. What are the wins that you need to happen within your workflow? There are likely tasks which will provide value day to day. Identifying these tasks will help guide your product and service selection. Remember that Artificial Intelligence is a journey that has no necessary endpoint. Like the digitization of the workforce, AI will gradually become a regular feature of the products we regularly use without realizing it. We need not try and find the most complicated application to create a strong AI infrastructure within our toolkit.
More value can be found through smaller, but stronger wins in the early stages of AI adoption. Think of this like winning lots of small pots during a poker tournament. By taking your time and playing your hand equity appropriately, you will gradually find yourself with the largest stack of chips. Rushing into a complicated AI initiative without sufficient planning is the equivalent of going all-in without looking at your cards.
Finally, it is important to ask, “so what?” Pay close attention to what the vendor is telling you when they are talking about their AI solution. You could easily be at risk of distraction with dazzling tricks and performance stats. It is critical to keep course and maintain the focus of the vendor. If the vendor starts talking about ROI’s, confirm that this is an ROI on the solution, not the product itself. If they do not understand the difference, then you may need to shop around a bit more. Be clear about exactly what you are looking for and define what you are not interested in hearing about. This will help both yourself and the vendor focus on what matters.