Henning Hatje
Henning Hatje

The path from buyer to bAIer

The path from buyer to bAIer

We want to look back at the DPC "From Buyer to bAIer" and share our insights with you!

For now, we would like to thank Philipp Müller and Konstantin Hokamp from Signatrix for the expert presentation and all participants for the insightful discussion. It was again great fun to discuss highly relevant topics in procurement in a controversial way.

Since artificial intelligence is expected to become part of many purchasing processes in the medium term, the path from buyer to bAIer was discussed. In addition to possible use cases for purchasing and specifics of AI procurement, the limitations that buyers encounter in AI projects were explained in particular.

We take away the following lessons learned from the event.

AI is already finding diverse applications in the ordering process optimization. 

In general, AI can be used where repetitive tasks occur. Examples in ordering process optimization are the definition of requirements or the supplier search. Based on historical data from the company, or from external public databases, an AI application can make suggestions and support the buyer in his decision. Consequently, the speed and quality of the procurement process is increased. This is reflected in cost and time savings.
In the long term, AI should also be able to make forecasts on price developments or supply chain risks. At the moment, however, this is still hindered by the difficulty of collecting the relevant data.

AI combines elements of a software & service capability that need to be considered in the procurement process 

AI is far more than just software. At the heart of an AI application is a set of trained data models. For example, they interpret images, transcribe voice recordings, or perform other complex tasks. Maintaining/training these data sets is costly and results in AI being a combination of a service and software.
The success of an AI application is therefore based on the one hand on the data models but also on the competence of the partner. Consequently, data quality and quantity is absolutely critical. However, it is problematic that data in companies is often disorganized, difficult to access and non-transparent. Furthermore, most AI projects fail due to a lack of communication in the initial phase. An ideal partner should therefore have clearly articulated expectations, talk about implementation issues from the start, and make pragmatic decisions. 

An insight into the testimonials from our participants AI combines elements of a software & service capability that need to be considered in the procurement process 

There is a lack of clarity around the term AI. For example, buyers are not always sure which concepts fall under AI and which do not. Finding a universally valid definition for AI is difficult. This problem derives, among other things, from the fact that the concept of intelligence is not clearly defined. In the discussion, Optical Character Recognition (OCR) fell as an example. Here it can be argued that OCR is only pure text recognition, while AI would be able to draw insights from the text. However, analogous to this example, there are many other cases where the boundaries of AI are not clearly formulated. The lack of conceptual ambiguity logically complicates the acquisition of AI.

Furthermore, the problem of data availability, already mentioned in the presentation, was confirmed in the open discussion. For many companies, this is the biggest hurdle to implementing AI solutions. It requires data backup systems in which data is backed up dynamically, can be restored at short notice and, above all, is always available. Again, this also means that data security standards have to be adapted. It can therefore already be deduced that establishing a high-performance data infrastructure is very resource-intensive. However, data availability is not only essential for AI, but in general also for future competitiveness.

Lastly, the scoping of AI projects was critically examined. Often, from the beginning, too high expectations are created for AI. Misleading statements lead to disappointing results and negative attitudes towards AI. Here, AI companies are also responsible for making statements that are too high when marketing their solutions. But the conceptual ambiguity of AI also plays a role, and prevents buyers from intervening early in the scoping process and from raising expectations too high.

We're excited to see how AI will evolve in companies (and especially procurement).
If you missed the DPC, and have questions or comments about the journey from buyer to bAIer, we're always happy to exchange ideas! And also, if you have any exciting purchasing topics that the DPC should definitely discuss, feel free to contact us anytime!

Don't miss the next webinar!

The DPC team is already working on the next event:

"Risk and Crisis Management from a Procurement and Supply Chain Management Perspective."

March 22 at 5pm. 

Register on the DPC website to stay up to date!

New call-to-action