Skip to main content
RISE logo

The path to AI: A getting started guide for businesses

AI can free up time, streamline and improve processes in sales, customer service/support, recruitment, product development, marketing... basically, anything in a business that is or will become fully or partly digitised. AI can even provide new business opportunities. But how do you get started? How do you succeed with an AI initiative?

It is no secret that global tech giants like Google, Amazon, Facebook and Spotify make billions with their data through the use of AI. But how can businesses that are not “tech first” improve their business through the use of AI and their own data?

– “Businesses that are not “tech first” often do not understand the data they have or its value,” says Fredrik Olsson, Senior Researcher at RISE, who helps businesses get started with AI and better utilise their data.

Data may encompass anything from web logs, machine logs or device logs to e-mails, support messages, chats or documents. Data can be used to free up time or streamline and improve processes – and even to generate new business.

In order to get started with a first AI project correctly and to minimise the risk of flushing money down the drain, it is essential to have an effective strategy in place.

– “Many get started with AI for the wrong reasons. An AI project is often started just for the sake of it, out of a fear of not keeping up and the starting point is usually the question “what can we do with our data?”.”

Take on a real challenge in the business

Step One: Identify problems that would be worth solving

The first step, which is perhaps somewhat contradictory, is to resist the temptation to start from the basis of the data you have access to.

– “Take on a real challenge in the business instead, so that the first AI project solves a real problem. Then it will be worth something and it will be more natural to assign a budget for the project. Ask questions such as: “What are the pain points in our business?,” “Where does it really hurt?,” “What is important for us to solve?”

Here are some examples:

  • Expensive support. Some support requests are handled by two or three people or are bounced back three or four times before the customer is happy.
  • Repetitive routine tasks. Each week an employee has to spend three hours looking for the right information to move forward.
  • Interruptions in production. There are interruptions in production at factories around the globe each day. There are huge amounts of logs relating to the issue but to analyse these to identify the underlying patterns of the issues would be an overwhelming task.
  • Product uncertainties. Which product should we produce next? Which has the greatest chance of success?
  • Loss of knowledge. Internal knowledge is not passed on. Who has knowledge of both 5G and energy issues? It takes a long time to connect the right people to the right knowledge internally.

– Feel free to assign an estimated price tag to each pain point. What does it cost in terms of money, working hours and loss of potential opportunities?

It is better to start small in order to minimise risk

Select a limited project

When you have completed your homework of identifying pain points, the work on finding a suitable first AI project can commence. Here, many businesses have sought the assistance of RISE.

– “Together, we will take a look at the list of pain points and try to find something that is both simple enough and limited enough to complete in 4-12 months. We want to avoid spending 24 months on a major project that may not be as good and that ends up costing several million Swedish kronor. It is better to start small in order to minimise risk. Even in a small project, it is possible to learn what you can do with your data, while also building mutual trust for future larger projects. Smaller projects are also easier to redirect if needed.”

Choosing a limited first project often requires a few meetings. The project must be allowed to take its time if you want to achieve specific project outcomes that have a practical benefit.

– “The project will often involve developing a web application of some sort or conducting a specific experiment using the company’s own data. We want the project to lead to something specific and the limitation is therefore crucial.”

Should more data be logged? Or another type of data?

Is there data available or does it have to be generated?

It is not only the limitation that takes time at the start of a new project.

A key variable that will influence the nature of a first project is the available data relating to the pain point you want to tackle. AI requires a large volume of data, in order to train the intelligence, draw the correct conclusions, generate relevant texts and graphs and perform various tasks.

– “What data is available and is it available in a format that can be analysed? As an example, chat logs and e-mail messages are easy to work with, while PDF files are often difficult for computers to understand. Should more data be logged? Or another type of data?”

During these initial discussions, we will also take a closer look at the various aspects of the problem to be solved.

– “Discussions about problem areas tend to be useful for businesses in and of themselves. They are forced to explore key issues that accelerate their business development.

These discussions also result in businesses asking, time and time again: “Does it work?,” “Is it better to do A or B?”.

– “The goal, of course, is to find a way forward in which you can find a good balance between benefit and effort.”

The most important thing is to get started

It takes work to implement AI

Putting AI to use also involves a commitment to provide the computer with feedback on an ongoing basis. When does the computer get it right? When is it wrong?

– “AI develops from the feedback it receives on what it delivers.  It is not the same as buying an Office package. It is an investment that requires maintenance.

For example, unexpected distortions of data may start to emerge, causing the AI to draw strange conclusions. A train derailment in Gothenburg may, for example, result in many incoming e-mail messages on the subject, but it does not necessarily mean that Gothenburg is more interesting to customers in general.

– “When you start working with AI, you need to take into account the fact that it might also require some effort.” We want to help businesses figure out what they can and cannot do with AI and their data. There is no better way to improve your AI skills than to get started on a small and specific first AI project.

– “The most common approach is that we start by optimising an existing flow. In connection with this process, businesses discover how they can start thinking more AI. You have to think carefully about the collection and processing of data and, as part of the bargain, you also become a better client and buyer. You start to realise that AI can allow you to redefine the way you work or even how to “disrupt” your niche or industry and create new types of offerings. The most important thing is to get started.


The Center for Applied AI at RISE carries out cutting-edge research in AI, connects expertise and applications within RISE, and explores a wide range of innovative applications with industry and the public sector.


Center for applied AI at RISE



Published: 2021-01-15