GenAI in Action – A Blueprint

According to experts like Sam Altman (CEO of OpenAI) and Steve Young (VP of Dell Technologies), 2025 will mark the year when AI projects transition from theory to implementation. The first AI agents capable of autonomously supporting decision-making processes could revolutionize daily work. One thing remains clear: successful AI solutions rely on trustworthy and high-quality data.

Learn more about the role of AI in driving a successful digital transformation in our whitepaper.

A key aspect of implementing GenAI is the ability to adapt flexibly to varying requirements and create solutions that deliver genuine value. Not every challenge is a "nail" that can be addressed with the "hammer" of GenAI.

In other words: GenAI is often seen as a universal tool to be applied to every use case—this frequently leads to confusion about where and how AI can be most effectively utilized. However, by specifically identifying use cases where GenAI creates real value, it becomes easier to put these solutions into practice. But what approach is best suited to identify the right GenAI projects?

To address this very question, the diva-e AI team has developed the "AI Project Blueprint." This systematic guide helps businesses successfully implement GenAI solutions. It is divided into key steps that every GenAI project should consider:

Strategy Development: AI should generally be part of a company’s overarching strategy. Key topics such as governance, data quality, and employee involvement are critical here. When designing specific GenAI solutions, two elements form the foundation for a successful approach:

• Use Case Definition: Successful projects begin with clearly defined use cases. Templates are useful for collecting, prioritizing, and evaluating ideas based on predefined frameworks.

• Measure KPIs: Whether it’s download figures, conversion rates, or efficiency gains, the right KPIs make the success of GenAI measurable. The rule of thumb: less is more. A maximum of three "smart" KPIs per use case is a good guideline. In addition to KPIs for model quality and system quality, the evaluation should primarily focus on KPIs that measure business impact.
  

Solution Design: Depending on requirements, there are three solution approaches: off-the-shelf tools, hybrid models, or custom AI developments. Hybrid models, which combine pre-built solutions with tailored technical implementation, are currently highly recommended. They offer a balanced mix of quick time-to-market and manageable costs.

Prototyping: Rapid experimentation is crucial to gain insights before extensive implementations. Testing LLMs (Large Language Models) early, for instance by running them locally and feeding them datasets, can help determine which use cases are truly viable. This allows for the creation of a "proof of concept" early in the process.

Implementation: Once prototypes have been successfully validated, the integration into existing systems takes place. This step is critical and requires careful attention to technical fundamentals, including interfaces and the systems affected.

Workforce Enablement: Ongoing employee training is essential. All stakeholders should be involved early and prepared for the use of AI and GenAI.

Learn more about how to implement agile collaboration strategically here.

One example of the successful use of GenAI is a global manufacturer in the energy technology sector. Together with diva-e, the company integrated GenAI into its download center to better support customers. With a dialog-based search function, customers can find manuals and product data sheets faster. The results: fewer customer service inquiries and increased efficiency.

Another marketplace project also highlighted the potential of GenAI: here, product descriptions are currently being standardized and improved through automated suggestions. This solution supported both internal teams and marketplace participants. In the medium term, there are plans to create an interface for marketplace participants to provide them with ready-made texts directly.

Find insights into business models and trends around marketplaces here.

Curious to learn more? In our Digital Bash presentation, we provide deeper insights into the world of our "GenAI Blueprint" and demonstrate, using the two examples mentioned above, how companies can leverage it effectively. (German only)

Fabian Huck has been with diva-e since 2019 as Senior Consultant and Services Director for the USA. His focus is on the strategic expansion of the US business and translating business requirements into tailored software solutions. As the primary contact for CMS and commerce projects, his areas of expertise include solution consulting, business analysis, as well as project and stakeholder management.

As part of the diva-e AI Hub, Fabian advises companies on artificial intelligence and supports them in successfully implementing technological transformations.

For further information on AI topics, visit the diva-e AI section. Our experts are happy to answer your questions and engage in professional discussions.

With over 70 events annually, all of which are free to attend, the Digital Bash is the largest web event series in the digital industry in the German-speaking region. Like our blog format edge, it is aimed at everyone working in the digital field: decision-makers and executives, software developers, marketing specialists, and e-commerce experts.