Horizontal Versus Vertical AI

Horizontal AI offers new opportunities for SAP users and presents companies with new challenges. Systems such as ChatGPT and Perplexity are no longer just gimmicks; they provide practical support in daily work, such as creating documents, evaluating large amounts of data, and automating communication. In the traditionally highly structured and process-oriented SAP environment, horizontal AI technologies can help make routine tasks more efficient. These systems are generally cloud-based, meaning Internet access is all that is needed to use AI-supported text, image, or analysis functions. Users can complete complex tasks, such as text creation, image generation, and answering complex questions, even without in-depth technical knowledge. These systems are typically used as part of transparent subscription models, making the costs easy to calculate.

Lack of specialization

However, despite their versatility, horizontal AI systems have limitations. Their greatest strength—their universality—is also their greatest limitation. This is because they usually lack specific industry knowledge or in-depth specialist knowledge. For companies with highly specialized requirements, horizontal AI solutions can quickly reach their limits. Additionally, there is a growing administrative burden. Anyone using several AI tools simultaneously must coordinate and control their use, which quickly leads to increased complexity. Data protection is a particularly critical issue. Entering sensitive information, such as customers' personal data, into unapproved AI applications can have serious consequences. Violations of the General Data Protection Regulation (GDPR) are legally risky and damage the company's image. Horizontal AI is a powerful tool that should be used with restraint. Its benefits are undisputed—provided the technology's limits are recognized and taken into account within clearly defined guidelines.

Example of vertical AI: Joule

Unlike generally applicable horizontal AI systems, vertical AI is designed for customized solutions in specific industries or company divisions. Based on company-specific knowledge, it draws on internal data to address specific challenges and improve industry-specific processes. The result is a more precise and targeted use of AI that aligns closely with real workflows and requirements.

Examples of vertical AI include Microsoft 365 Copilot and SAP Joule. Microsoft's solution is deeply integrated into the familiar Office environment, offering intelligent functions in Word, Excel, and PowerPoint. Copilot provides text suggestions, automated data analyses, and presentation creation based on internal content. It understands the individual processes of a company and offers support where employees work daily. SAP Joule takes it a step further. As an AI component within the SAP world, it analyzes internal business processes, recognizes patterns, and provides data-supported recommendations. The result is an intelligent assistance system that accelerates decision-making, deploys resources more efficiently, and measurably improves a company's overall performance.

The strengths of vertical AI lie primarily in its depth of specialized knowledge. These systems don't provide general answers; they rely on specific knowledge. This results in highly relevant solutions that are faster and higher quality than universal AI models. Customized functionality also ensures greater integration into existing work processes and higher user acceptance. However, developing such systems is complex and requires in-depth knowledge of the subject matter, close coordination with relevant departments, and robust technical infrastructure. Integrating them into existing IT landscapes can also be difficult, especially if processes are highly individualized or legacy-oriented.

Conclusion

The choice between horizontal and vertical AI systems depends on a company's specific requirements and goals. Horizontal AI systems are versatile and easily accessible, while vertical AI systems offer customized solutions with specialized functions and knowledge. However, the safe use of AI requires careful consideration of legal requirements, particularly the EU AI Act. This act ensures that AI systems protect fundamental rights and comply with security and data protection requirements.

By implementing a well-thought-out AI strategy and considering the various benefits and challenges, companies can effectively use the transformative technology of AI to improve their processes and gain a competitive advantage.

To the partner entry: