By Christian Setzwein February 24, 2021
As a product manager, you are eager to advance your company and have come across an opportunity to try your hand at artificial intelligence. Reading up on the subject in magazines and blog articles has already sparked some initial ideas, and the initial feedback from your team has also been positive. Now you are ready to get started – but how can you best begin this journey?
The hurdles to getting started
As with other projects, as soon as you start dealing with a topic in more detail, you will inevitably encounter hurdles and obstacles that will at least delay the start of the journey, if not make the whole journey an unpleasant experience that you would rather quickly forget. But stop! – I’d like to help you avoid this. It’s good to know the hurdles in advance, then they lose their terror and you can overcome them step by step.
So here we go: the hurdles I’ve encountered so far in companies when getting started are as follows:
There is a lot of familiarization work to be done in order to understand what AI actually is and how exactly the machines learn. There seem to be an incredible number of possibilities (use cases) in which machine learning can be used. It takes a lot of time to get an overview of all of them. Finding the right areas of application in the company requires a team from different professions and different parts of the company. Knowledge about machine learning is unevenly distributed within the company. The data analysts are very far ahead, while many others are lagging behind and would first have to be brought up to a common level of knowledge. There are concerns that the balance of power within the company is shifting: Is the data analytics department now gaining power and we product managers have no say at all anymore?
- It is very difficult to bring together the necessary knowledge in the business area and the necessary knowledge in the area of machine learning (statistics, probabilities, algorithms).
We have already overcome these hurdles
And now, please, everyone stop and think. Does this look familiar? Haven’t we already been here before, facing similar problems? Many years ago, some of us still remember, the opportunity arose to program computers to solve business problems. Even then, those in charge of the business did not have to understand how to program computers themselves. Looking back, the hurdles mentioned were gradually overcome by the following solutions:
- The development of a new solution takes place in a team. In the beginning, I still remember, it was a business manager and a developer (someone who could program, usually self-taught) who developed a program together. Later, the teams grew larger, each person in the team had specific tasks such as describing business problems, writing programs, testing programs and operating programs. Everyone in the team understands something in their area, but the overall solution will only be good if everyone works together.
- Power balances were always decided in favor of the business side. It is not the programmers who have the power (although they are extremely important and a rare commodity), but rather the business managers seek the best solution for the customer.
- Everyone acquires the necessary knowledge in their profession, so it is not too much work for individuals. Together, there is enough knowledge to use the new technology. If a company is missing a knowledge component at the beginning, this can be compensated for by external specialists. The important thing is that the knowledge is then learned and built up internally.
Once you have considered all these parallels and taken courage, the journey can begin.
Find the right use case - the AI Innovation Sprint
Starting somewhere is one option. In my view, however, it is better to search systematically in the shortest possible time. A format that is used at Google for product development is useful here: the Google Design Sprint (see below for a recommended reading). Based on ideas from Jake Knapp and his team, I have developed the AI Innovation Sprint for my consulting practice in the introduction of AI solutions.
The AI Innovation Sprint enables a team in just five days**
- to build up shared knowledge about machine learning and AI.
- to discover and evaluate as many possible areas of application for their own company.
- to build prototypes for selected solutions and then validate them with customers or colleagues.
As with the Google Design Sprint, the whole process is divided into five steps on five different days:
Understand: Understanding AI as a technology and the fields of application for AI Ideate: Collecting ideas for your own use cases Decide: Understanding the ideas in more detail and then deciding on three candidates for implementation Prototype: Developing paper prototypes for the three candidates (application scenario, learning environment)
- Validate: Validate the solutions with customers, prospects and colleagues. Then select the candidate with whom to start the first AI project.
The advantage of the AI Innovation Sprint is that it is strategically anchored, very systematic and delivers results quickly. The first project can start and a team is ready to manage the implementation in the company. A lot of feedback can already be built on this.
In my AI Innovation Sprint Workshops, I then rely on a specially developed AI Canvas, which, based on the questions asked there, makes the search process easier for the team and also ensures that nothing is simply forgotten. As a methodological basis, I rely on the [Liberating Structures](https://www.amazon.de/Surprising-Power-Liberating-Structures-Innovation/dp/0615975305/ref=sr_1_1? __mk_de_DE=%C3%85M%C3%85%C5%BD%C3%95%C3%91&dchild=1&keywords=Liberating+Structures&qid=1610729964&sr=8-1) to promote innovative thinking and team spirit among the participants.
Courage, taking the first step is important!
I wish you much joy and success as you and your team begin the journey and take the first steps. As with all topics, it is important to get started in order to gain your own experience. There is no need to worry if not everything is ready, you can also buy missing things along the way. What is important, however, is to know the direction and to have a travel plan, such as the 5 stages of the AI Innovation Sprint. Just starting out is also possible in case of doubt – but for me, the possible success would then be too random. If you still have questions about the format or would like external support for the first steps of your journey, please feel free to contact us.
As always, suggestions are very welcome.
Recommended reading:
Knapp, Jake et al. (2016): Sprint: How to Solve Problems and Test New Ideas in Just Five Days, Redline Wirtschaft
Post Image
