Most AI experts believe that a use case identification workshop is a critical starting point for companies in their AI journey. After carrying out several of these workshops, I personally second that notion. Nonetheless, the different components of a workshop must be adequate for a valuable outcome.
In this post, I will define what I consider to be the three principal components that any workshop should include – AI formation, business vs. technical expertise, and the right prioritization attributes.
It is impossible to anticipate a favorable outcome in an AI use case identification workshop if the participants do not possess a certain level of AI knowledgeability.
Like mathematics, software engineering, navigation, or astronomy, artificial intelligence is in itself a discipline with its own set of concepts, frameworks, and peculiarities that need to be understood to be appropriately applied. What do you suppose would be the outcome of starting sailing around without understanding what tiller, mooring, tacking, or jibing means? Unhopeful. Luckily, there is no risk of collision in AI use case identification workshops. Admittedly a formation will make absolute sense.
Imagine the participants being capable of differentiating between structured and unstructured data and identifying the distinct data types within their day-to-day activities. Also, visualize the inclination of the participants to roughly understand the different AI capabilities such as computer vision or natural language processing. Furthermore, you can imagine them correlating the types of data and the AI capabilities with the processes in their business units. Now picture what the workshop’s outcome would look like if the participants did not know any of that knowledge…
The formation should depend on its length, how technical the workshop should be (depending on your target group), and, naturally, what your resources are. For instance, in Germany, it might be acceptable to have a lengthier workshop with formation included, whereas you might need a more minimal approach in the US. Also, suppose your target group is already AI experts. In that case, the workshop could go down into more technicalities when prioritizing use cases, whereas, with non-AI experts, you must stay at a higher level. Therefore companies also prefer to offer the AI formation as a separate service from the workshop. This, respectively, means more resources are needed. At AI Shepherds, we have designed a unique AI use case identification workshop where the formation is targeted to a non-technical audience.
Business vs. technical expertise for proper identification
The second step in an AI use case identification workshop is to carry out the actual identification. Combining the business domain and AI technical expertise for a meaningful outcome is mandatory.
The standard practice in this phase is to select a business unit to focus the identification of use cases. Next, the business expert and the AI technical expert will go through the processes within that unit while identifying wherever AI could play a role in it. Envision for a moment that the business unit is “human resources,” and more specifically, the “recruiting process.” During the workshop, an HR employee with several years of experience in the company knows the process and is an existing internal or external AI expert. Both experts go through each step of the process identifying the data generated, used, and processed, the inputs and outputs, the activities done and how long they take to be completed, and the repetitive tasks. For example, once the two experts get to the process’ exercise where the recruiters go through each CV manually, the AI expert proposes the possibility to automatically and intelligently scan CVs to identify the most suitable candidates (common AI use cases in the HR field). The AI expert can ask the right questions regarding the technical requirements to make the use case feasible, whereas the business expert can picture the exact context and needs to be covered by that use case.
I’ve been involved personally on numerous occasions during these sorts of face-to-face business vs. technical discussions. From my experience, not having the proper people in place during the workshop will diminish the quality of the outcome.
Choose your best method for prioritization
Having the proper formation and experts will result in a list of use cases. All that is left is the third and last component for an appropriate AI use case identification workshop – the use case prioritization.
There are different best practices when it comes to prioritizing use cases. Nonetheless, they all have one thing in common – stay at a high level. To do things both agile and efficiently, it is inadvisable to avoid deep-diving into how much the revenues would increase, the costs decrease, or how much faster a process would be because of a use case. It is often impossible to assess due to the intrinsic nature of data science. Moreover, a workshop is where the idea is to have a simple approximation of what use cases would give the insight to develop first. Therefore, using high level weights/scores (e.g.: 1 = low; 5 = high) is the ideal approach to go by.
Attributes to measure during the prioritization phase can be:
- Measurable attributes: which include cost, revenues, customer service, and manufacturing temperature.
- Strategic attributes: synergies of a use case with vision, the use case’s impact on a specific goal, etc.
- Capability attributes: the required infrastructure, skills, cultural implications, etc.
Companies can define standard attributes to assess during the prioritization phase. However, each business unit is unique, which accordingly, the need for additional features should be considered. Once each use case has been scored, it is then possible to prioritize them.
There is no exact formula for structuring an AI use case identification workshop. It will regularly depend on the need, context, and target required of the workshop. Based on what we see in the market, our recommendation is to ensure the participants have received an AI formation. We also recommend that business and AI experts be present during the identification phase and choose the right attributes during the prioritization phase.