There have never been so many innovations as there are now

What are real innovations

Where we have mainly built software for the last 20 years that simply maps existing, manual processes in structured software, today the innovation is taking place in completely different areas.

Today we develop software that changes entire processes or business models based on data. There are 5 different approaches:

1. Benchmarking: improvements based on comparisons and benchmarking. As there are almost exclusively multi-tenant SaaS applications today, anonymous data from all customers can be used to compare certain KPIs and, on this basis, to offer benchmark data to every company. For the customer, this has the advantage that you not only get best practice from the provider, but also information on best performance. This then serves as the basis for optimization measures according to the motto "if the other company can do it, then we can do it too".
2. Usability: improvement of the usability of software because the software already knows what the user wants to enter based on previous processes or which data is relevant in the current process step. This sounds banal, but it has major implications in practice. For example, there are vendors who can automatically select the right product group based on article texts, which in turn automatically assigns the right suppliers and templates. This not only simplifies the process, but also creates a much better data quality, since users often choose the first entry for the sake of simplicity when making manual selections, which means that all the evaluations and reports are wrong.
3. Recommendations: the generation of recommendations and their automatic execution changes the purchasing and also the software landscape in its basic features. The development of recommendations has the difficulty that you not only need good developers and data analysts for this, but also specialists who can make the decisions today and articulate exactly on which data basis and which empirical values they make these decisions. The problem with the development of recommendation engines is, on the one hand, that the product managers and developers of the providers have no practical procurement experience and, on the other hand, that in most cases not all the data is available at all to be able to generate really clever recommendations. Nevertheless, there are already initial approaches to generating recommendations in different areas and offering them to customers.
4. Guidance in choosing the right software features is another area where companies are making great strides. As with ChatGPT or Microsoft Copilot, the aim is to give the computer instructions without knowing which menu item is relevant for this, and the software understands the request, translates it into commands, executes them and returns the result. Systems deliver real added value here that not only deliver static texts (as they were in the manual in the past, for example) but can also deliver dynamic content from the database, such as the current spend with a supplier in a certain period of time.
5. AI agents: since 2025, the most important trend. Instead of supporting a user, agents carry out complete tasks on their own: they process a purchase request, send RFQs, compare quotes, negotiate within defined limits or chase a missing order confirmation. The buyer defines the rules and approves the result. Almost all established suites now announce agents (for example SAP with Joule in SAP Ariba), and a new generation of AI-native startups such as Lio, Didero, Pivot or ORO Labs raised large rounds in 2026 on this promise.

Where are the differences between the software providers

In practice, the innovative possibilities of a software vendor depend on several factors:

1. Development environment: anyone who is familiar with it knows that in the last 20 years, and especially in recent years, many new development frameworks have come onto the market, with which new functions and features can be built very quickly. This not only applies to the development language but also to the database architecture. SAP has taken a big step here with S/4HANA and is now rebuilding SAP Ariba around AI, but this is only a small part of what is possible today. Choosing the right database for the respective requirement is not only more important today, but also more complex than ever before. Now, most of the large, established players today use development environments that are more than 10 years old and present a consequent obstacle to the rapid development of new features or the improvement of existing functions using machine learning. This is a clear disadvantage for some big players.
2. Existing data: every innovation that is also based on machine learning requires a lot of data with which an algorithm can be checked and improved. These large amounts of data are only available to very large corporations or new startups that cooperate very closely with large enterprises and are allowed to use their data to improve their algorithms. In general, however, established players have an advantage here because they can access data from different customers from different industries in order to improve the quality of the algorithms.
3. Ownership structure: the ownership structure is so important because different owners look at different investment periods. Owner-managed companies or listed companies have to achieve their quarterly and annual targets, but can also start developments that only really pay off and revolutionize the market after 5 years or more. Private equity companies have an investment period of 3 to 6 years, which means that an investment must pay for itself after a maximum of 3 years. If you pursue long-term strategies, you can completely redevelop functionalities from acquisitions and integrate them into your own source code, which can significantly reduce the complexity. Of course, this also depends on the number of M&A activities, since this is of course no longer possible with many large acquisitions. SAP customers know the problem very well, since many applications that SAP has bought in recent years only communicate poorly with each other.