The "Purchasing" function (including S2P) is on the verge of a new transformation. Its positioning and contribution will be disrupted in the coming years by the new digital revolutions.
Technologies such as Artificial Intelligence (AI) or Big Data are entering a phase of relative maturity and their implementation is now much more "tangible".
Machine Learning" or "Natural Language Processing" are undergoing significant development and their integration is becoming increasingly easy (especially through APIs).
We are convinced, and our initial feedback confirms this, that we have entered a new phase of development of the Purchasing function. New tools and new uses will force decision-makers (starting with Purchasing Directors) to rethink their organizational and governance structures, as well as the objectives assigned to the Purchasing function.
While the "traditional" sources of productivity have been exhausted, the need to refocus the actors of the "Purchasing" process on tasks that create added value remains entirely. The good news is that the actors (buyers, managers, management controllers, suppliers, etc.) spend more or less half of their time on repetitive and low added-value tasks (data processing, information retrieval, writing, etc.). It is now possible to automate them via protocols ("case management").
Obviously, the objective is to allow Procurement professionals to focus on complex cases (or on handling exceptions) and to spend more time working with internal customers (especially before the procurement process) and suppliers (after the procurement process).
The Purchasing function uses and produces a large amount of internal and external data (specifications, contracts, suppliers, non-financial ratings, article catalogs, orders, stock entries and removals, cost and budget accounting, etc.), which is perfect for an AI revolution. Moreover, the rise of APIs and ETLs offers new possibilities for exploiting this data and leads to the development of AI solutions.
The Purchasing function can now make better decisions very quickly, based on information collected, analyzed and optimized using algorithms. These are fed by the data produced "in house" but also by those available (purchased or not) and integrated via an API system. The analyses carried out are not only descriptive (what is happening) but also predictive (what will happen). They can also be deductive and suggest to the user possible actions in relation to a given situation.
The AI will modify the "business" actions but above all the positioning and contribution of the purchasing function. Some tasks will disappear, others will be significantly "increased", others will be created. Typically, we anticipate a strong development of the "Category Management" and "Purchasing Management Control" functions.
Among the use cases we identified, one is quite representative of the direction we are taking. Machine Learning" and "Natural Language Processing (NLP)" significantly increase the process of drafting and monitoring contracts.
The automation of these tasks results in considerable time and financial savings.
These tools and algorithms, which we are developing, are integrated into existing IS and improve over time through successive loops but also through ever-increasing access to data.
Faced with the universe of possibilities, but also the scale of the transformation project, organizations will have to prioritize the tasks they want to automate.
It seems unlikely that an AI solution will cover, in the short or medium term, all the sources of automation in the S2P process.
We recommend a step-by-step approach to adopt AI tools. It is obviously preferable to start its transformation towards AI with "simple" projects, generating quick gains and understood by all.
An AI project must be conducted along three axes:
To manage our AI projects, we have implemented a real purchasing process and selected several technologies that we believe offer the best agility and durability. Given the potential they offer, the cases of use retained by our Clients and the change management issues they face, we only exploit the small portion of everything they would allow us to do.
The possibilities for AI to "increase" the purchasing function are numerous and obvious. The question is no longer so much about the opportunity to move forward as it is about "How to move forward? »