How to Use Case - the systematised development of (gen)AI ideas using a lightweight use case framework

In the energy industry, the integration of artificial intelligence (AI) is a key component for future innovations. In particular, the integration of generative AI (GenAI) opens up new opportunities to increase efficiency and overcome complex challenges. In order to develop (Gen)AI topics and use cases successfully and in a targeted manner, a structured framework is essential. In this blog post, I describe a multi-stage framework that provides guidance for the development of use cases and takes genAI-specific requirements into account.

1. Discovery - finding the topic

The first stage of the framework is known as discovery. It focuses on the initial identification of potential genAI topics and use cases in all dimensions and scales. Various sources are used for this, including trends, market developments and internal requirements. When considering internal and external influences, particular attention should be paid to the aspect of interdisciplinary collaboration: Different perspectives and expertise enable the identification of diverse topics. The establishment of a dedicated genAI competence centre can help to support and coordinate the generation of topics and also raise awareness and educate people within the organisation about the still poorly understood topic of "genAI".

2. Steering committee

A steering committee is set up to ensure the strategic direction of the framework and subject the initially identified genAI topics and use cases to an initial validation. The initially assessed potentials and their fit with the (gen)AI vision/strategy play a decisive role as evaluation criteria. In addition, the first potential team constellations are agreed here, which are dedicated to the identified topics in the further course of use case development. When selecting the members of the steering committee, data protection officers and works councils should also be included alongside genAI experts, IT and management in order to address the genAI-sensitive aspects of "security" and "data protection" at an early stage.

3. Generation of ideas

Once the initial genAI topics and use cases have been identified and selected for further investigation, it is important to concretise the understanding of these topics. As dimension and scope can vary greatly, it may be necessary to perform a topic and cluster analysis to identify dedicated and manageable use cases. Once this has been done, previously defined interdisciplinary teams generate solution ideas for the genAI cases using creativity techniques.

4. Evaluation and prioritisation of ideas

The genAI use cases are evaluated and prioritised by means of a thorough validation based on previously defined criteria in the core dimensions of "complexity" and "business benefit". The business benefit is significantly influenced by, for example, customer experience, productivity/efficiency gains, cost savings or image gains. Complexity, on the other hand, includes criteria such as implementation effort, data quality/protection, dependencies, required skills and stakeholders involved. T-shirt sizes (e.g. S, M, L, XL) are a tried and tested means of classifying complexity. The weighting of the evaluation criteria in advance is particularly important in the genAI context. The primarily genAI-critical criteria of security, data protection and existing skills/competences should be emphasised here.

The evaluation of the genAI use cases is followed by a clear visualisation of the results. This can be done with the help of a prioritisation matrix. This creates the necessary basis for decision-making in order to prioritise the solution ideas against each other and also to identify possible clusters. The final prioritisation step involves transferring the use cases with the solution ideas into an implementation roadmap in order to take the implementation timeframe into account.

5. Prototyping

This is where the implementation of the previously evaluated genAI use cases begins. Initial prototypes are developed and tested for selected use cases. When developing the prototypes, genAI-specific requirements such as transparency, traceability and security must be (further) considered. For example, all prototypes must ensure that the decisions of the AI algorithms are traceable and that generative data is treated securely and confidentially. The aspect of AI hallucinations must also be checked during prototyping.

6. Implementation/scaling

Once prototyping has been successfully completed, the prototypes are implemented in the relevant areas of the company and, if successful, scaled and optimised. As genAI use cases can give rise to completely new applications and services, it is essential to create lasting trust and acceptance within the organisation through dedicated change management measures to accompany implementation. The aforementioned AI Competence Centre could also play an important role here.

Conclusion

In summary, the lightweight use case framework presented here provides valuable guidance for the systematised development of (Gen)AI cases in the energy industry. It illustrates the steps from topic identification to implementation and emphasises the importance of interdisciplinary collaboration as well as genAI-specific requirements such as security, data protection and competence development within the overall framework. Thanks to the structured approach, the potential of genAI can be tapped efficiently and anchored in the organisation in a targeted manner.

You can find more exciting topics from the world of adesso in our previous blog posts.

Also interesting:

• GenAI goes Energy - Use cases for GenAI in all areas and market roles