Large Language Models (LLMs) support strategic procurement with high efficiency by creating and improving texts, performing web research, or assisting in the use of software solutions. Mathematical and statistical AI offers enormous savings potentials that, compared to an LLM, are not as easy and self-explanatory to implement.<\/p>\n
A perfectly structured information system like WebCIS 4.0\u2014which enables reporting, benchmarking, and strategic purchasing decisions at the push of a button\u2014is an excellent foundation for analyses performed by integrated AI models.<\/p>\n
Market comparisons using price indices, detecting illogical price and quantity structures in tiered pricing, AI-generated suggestions based on learned patterns, and forecasts derived from learning processes are methods that would be unimaginable without AI.<\/p>\n
From our experience, those interested in mathematical AI models fall into two groups: opportunity seekers who immediately perceive the results of AI models as added progress in their work environment, and skeptics who criticize insufficient perfection in the assignment of market indices and reject comparisons of AI results with their own numerical world. The fact is\u2014without denying the errors of AI models\u2014never before has it been possible to obtain suggestions, ideas, and potential improvements at such speed, providing an expanded perspective on existing thought patterns. Fact-based rather than gut-feeling.<\/p>\n
Market Index Comparisons as Benchmarking at the Push of a Button<\/strong><\/p>\n When evaluating the performance of successful purchasing activities, one key question arises: Have material cost developments per product category been better or worse compared to market indices? Through AI-assisted assignment of product groups to the appropriate indices\u2014NACE, Eurostat, Federal Statistical Office\u2014not only selective comparisons but all product groups can be classified at the push of a button, including drill-downs to suppliers and materials. This creates an impressive basis for deeper analyses.<\/p>\n Homogeneity Index as a Measure of Product Group Structuring<\/strong><\/p>\n The index comparisons above naturally provoke doubts about the meaningfulness of existing product group assignments. AI-based homogeneity indices often show only low levels of consistency among the underlying comparison features within product groups. A compelling way to broaden one\u2019s perspective would be to introduce an additional alternative system with the sole aim of achieving the highest possible homogeneity index within product groups. AI models provide valuable suggestions here.<\/p>\n Multiple Regression Analyses to Identify Cost Drivers<\/strong><\/p>\n How do the characteristics of a purchased part influence its price? One option would be a time-consuming cost breakdown. But how many parts can be analyzed in a reasonable amount of time? And even then, internal discussions can be extensive depending on the outcome. A multiple regression analysis retrieves characteristics of materials\u2014weight, size, material type, manufacturing processes, and other features influencing price. The regression equation shows which influencing factor contributes what share to price formation and where there is room for negotiation.<\/p>\n Tiered Pricing: Logical Design or Coincidence?<\/strong><\/p>\n Lot-size-dependent prices remain a common method for handling planning fluctuations. But according to what rules are tier quantities and tier prices actually defined? Rarely is there a plausible explanation to this simple question. Two essential questions arise when defining tier prices: Which materials are even suitable for tiered pricing, and are the tiers set up such that fixed and variable costs enable logical price calculations for each tier? AI models reveal possible inconsistencies and solutions at the push of a button.<\/p>\n In the next step, actual tiers should be compared with potentially more economical tier prices. Considering the price advantages, and despite higher capital commitment, savings of 5\u201310% are realistic.<\/p>\n Does Increasing Purchasing Volume Influence Purchasing Results?<\/strong><\/p>\n What is commonly referred to as quantity or volume discount also exists in industrial procurement. Fixed-cost degression, improved planning reliability, and freight optimization often accompany rising volumes. It is surprising how often our system uncovers volume increases with barely noticeable cost reductions or missing bonus agreements.<\/p>\n In WebCIS, a single click is enough to identify potential suppliers with whom bonus tiers may be feasible.<\/p>\n All these examples illustrate how mathematical and statistical models of artificial intelligence already represent an essential resource in the procurement of the future\u2014and will continue to mature.<\/p>\n Werner G\u00fcntner (CEO)<\/strong><\/p>\n","protected":false},"excerpt":{"rendered":" Large Language Models (LLMs) support strategic procurement with high efficiency […]<\/p>\n","protected":false},"author":3,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-3181","post","type-post","status-publish","format-standard","hentry","category-nicht-kategorisiert"],"yoast_head":"\n