Making an appointment for a haircut, forecasting the demands of customers or even driving a car; as the performance and applicability of AI further increases, more and more organizations are willing to follow big tech giants by implementing AI in their processes and services. Our experience has shown us that the approach on implementing AI has an enormous impact on its success and sustainability, which leaves us with two to be answered questions:
- How to start off with AI?
- How to select a first use case?
How to start your AI journey? – Create a joint understanding of AI
When it comes to AI, buzzwords such as machine learning, deep learning, and big data pop into one’s head. Although there is a general understanding of AI – a machine with cognition – most stakeholders are not fully aware of the differences between involved technologies as well as their capabilities. That’s why the first step on your journey should be to create a joint understanding of AI among your stakeholders. Since AI embeds many components and characteristics and therefore a wide range of definitions, a simple model to create a common ground should be used. We at Capgemini made good experiences by clustering the technologies in our “5 senses of AI” model.
- Interaction (listen / talk, read / write): Is the ability to engage in a conversation with the user. The aim is to ensure that the AI component responds accordingly, and the interaction feels intuitive. Chatbots and voice assistants are common examples.
- Monitor (watch): Is the feature to capture and record unstructured visual input for knowledge creation.
- Knowledge (remember): Is about being able to store knowledge as well as effectively retrieving knowledge using databases and search engines.
- Analyze (think): Is the ability to recognize patterns and trends. It applies algorithms to knowledge to determine appropriate action or predict future consequences. It is the brain of the AI where the learning takes place.
- Service (act): This area uses technology to perform concrete actions depending on the input. Frequently used examples are virtual robots (RPA) as well as process mining.
Instead of seeing AI as a single solution, it should rather be seen as a combination of technologies that generate a solution. Just like humans use different sets of cognitive abilities to solve a problem, AI uses various technologies, e.g. machine learning, to fulfill a task. Depending on the desired outcome, different technologies, each resembling a cognitive function, are combined.
There are several options how to make AI tangible within your organization like embed it into your overall digitalization as one component for automation or to start with a first MVP (Minimum Viable Product) of a specific use case to get a better feeling of the potential of AI and embed it afterward. In the next step, we will show you what we consider when selecting a use case for a first MVP.
How to select a first use case? – MVP as a flagship project
After creating a joint understanding of AI with your stakeholders, it is time to start the important process of use case identification. This step on your journey is crucial for the successful implementation of AI, since the use case will act as a flagship project and therefore strongly influences the perception and acceptance of AI within your organization. Based on our implementation experiences we recommend following these guiding questions:
Does the use case generate a concrete value for your stakeholder?
The first impression counts. If there is no value for your stakeholder, interest in AI will inevitably decrease.
Is the process suitable for a first showcase?
The stakeholder should experience the impact of AI and in the best case it will cause a “wow” effect within your organization. The MVP acts as a door opener for future implementations.
How complex is the use case?
Always keep in mind that complex cases need more time, budget and resources for implementation and additionally bear a higher risk of failure. Try to keep the right balance between benefits and complexity.
Do you have a sufficient database?
A consistent and sufficient database is key for the successful implementation of AI, so make sure data is available.
Do you already have internal knowledge regarding this topic and are your SMEs available?
Existing knowledge and know-how boosts the progress of the project, especially in the early stages.
Does the use case fit into your group strategy?
Rather than being a stand-alone solution, AI should fit into your group strategy to promote acceptance and progress.
Just like AI is a complex field in computer sciences, its implementation within an organization is a complex task too, that must take many aspects into account. If you are at the beginning of your Intelligent Automation journey, we recommend starting by defining AI with your stakeholders – create a framework upon which technologies can be classified and the big picture of AI can be illustrated. Once a joint understanding is reached you can move forward to selecting the first use case following Capgemini’s guiding questions. Keep in mind that your further journey in AI substantially depends on the success of the MVP.