A brief review:

In Part 1, it became clear that AI can develop very impressive capabilities when given clearly defined tasks.

Using the software "Analytical Process Automation" (APA) - Alteryx, a neural network (an algorithm in AI) was trained and implemented in Part 2 in such a way that it can subsequently be easily applied by business users.

In this application in Part 3, it became clear that AI and especially neural networks are only good at the things that have been learned in training. However, if data occurs that was not learned in training, neural networks can make very unpredictable errors. Even errors that we as humans would not necessarily consider logical, because neural networks cannot analyze data the way humans can.

The last article, Part 4, dealt with the question: How can AI projects be implemented in SMEs? Using a process flow, the various phases of a data science project were presented.

With all these insights, one question remains unanswered: Where exactly can you use artificial intelligence as an SME?

Well, to be honest, there are a lot of different options that really depend on the processes and available data in your SME. That said, we can narrow it down to two main streams of AI applications. Each of these streams is illustrated by a few example use cases.

For SMEs, two areas are particularly prominent - automation and prediction.

Forecasting (Vorhersage)

AI in the field of prediction deals with statements about the future based on past experience. Some examples of this application:

"A local electric vehicle rental company wants to expand its fleet and is therefore interested in when peak utilization will be reached. AI here can develop a trend for the future from past data and make predictions about when and how high utilization will be."

"A wine supplier has put together a new product range. Customers should receive targeted recommendations for products they like best from the new range.  AI can infer areas of interest for groups of customers based on past buying behavior, making predictions about interest in the new products."

"A telephone consulting company has a long list of questions that are asked of each customer on the phone to match them to the right advice. This process should now be shortened. AI can use historical data on questions and answers and subsequent classification. This allows the AI model to dynamically recommend which question to ask next, depending on the last answer, to get a good and quick assessment of the customer without having to query the entire catalog."

Forecasting is based on modeling a process, and in this modeling, time can then be "fast-forwarded" to see the future state of the process. Based on this modeled future state, a forecast is made. It is important to realize that a model can never represent 100% of reality and therefore cannot make a 100% reliable prediction about the future. However, the more that is known about the process, the more accurate the model becomes. As described in Part 2, data is critical to AI performance, so the same is true in this case: the more high-quality data, the more accurate the prediction.

Automation

The application in automation can be summarized in a rule of thumb: "AI should be considered where digital tasks are repetitive.

Some examples:

"A company receives standardized multiple-choice forms. Employees read the forms and enter the answers into a database. With AI, this process can be automated. Using image recognition, marked fields are recognized and the result is automatically transferred to the database."

"With the webstore provider offering a new contact service, the store is receiving many more written exchange requests via email. AI can help here by presorting emails by topic and then delivering content only to employees who specialize in that topic."        

Automation can be used in many areas when very standardized processes are in place: Often in combination with image recognition, automated reading of printed text and forms, analyzing text content, extracting text content and transferring information to internal forms or databases. Standardizing text- and repetition-heavy processes can provide an interesting business advantage.

Inspired? - Are you looking to leverage the benefits of AI for your SME?

Are there process steps that could also fit into the AI space? Or is there a prediction that might have always been interesting? But is it really the right use case?

We at Banian AG are happy to help you with an initial consultation. Contact us by phone +41 (0)61 551 00 12, per Mail or via LinkedIn. And if the business question goes beyond AI, we are happy to help as a digitization partner from strategy to information management to data & analytics.