“In my fall 2004 analyses of a study on the use of web-based applications I made the following observation:
It is my position that a true centralization of procurement objectives requires a decentralized architecture based on the real-world operating attributes of all transactional stakeholders, starting at the local or regional level. In other words, your organization gains control of its spend environment by relinquishing centralized functional control in favor of operational efficiencies originating on the front line. This is the cornerstone of agent-based modeling.
So while the ISM, CAPS and AT Kearney report is just now beginning to acknowledge the emergence of this dramatic shift, their ability to look beyond the conceptual level is significantly hindered by their respective technological blinders. This is reflected in their assertion that the next decade will be dominated by center-led organizations.”
In the follow-up post to the one above, I wrote the following:
“Throughout the research period (which was partly funded by the Government of Canada’s Scientific Research and Experimental Development – SR&ED program), we consistently looked for ways in which a buyer could reliably procure commodities on a real-time basis outside of the confines of a centrally negotiated contract. To do this effectively, the buyer must simultaneously engage critical stakeholders such as suppliers, courier companies and customs brokers, e.g., the Metaprise. Given that off-contract procurement was at epidemic levels (which negatively impacted both the buying organization and an increasingly skeptical supply base), it was essential to lessen rather than increase the purchasing cycle time. This cycle compression was a critical component in that we wanted to eliminate the compliance issues that had plagued so many initiatives (as it still does today). In August 2003, the new technology successfully went live in a production environment for the DND. In this test case, the public sector organization realized a year-over-year 23% cost of goods savings for seven consecutive years while simultaneously reducing the number of buyers required to manage the contract to 3 from an original 23. Delivery performance and product quality also improved dramatically.)”
“Do you see the seemingly subtle yet critical differences between using a technology-led Agentic AI model and a human-led Agent-based model?
Suppose we as an industry don’t and continue to lead using an Agentic AI model. In that case, we will repeat the same pattern of technology-led implementation outcomes and have the same discussion regarding initiative failure rates 10 years from now as we are having today.”
Watch the following video from Microsoft CEO Satya Nadella within the context of everything above. My preliminary takeaway is simple: ERP, SaaS, AI, Generative AI, Agentic AI and anything else we develop tech-wise in the future outside of an Agent-Based model WILL ALWAYS FAIL!
I will use two words here to demonstrate why the CEO of a “technology company” may very well lead us further down the wrong road: Strand Commonality. Strand Commonality and its subsequent data output require the right “kind of human intervention” – otherwise, you will never get the correct data. Unfortunately, most humans are unable to effectively and consistently perform the kind of human intervention that is needed. I will delve more into this in future posts.
Procurement Insights 
Microsoft CEO Satya Nadella Talks About Agents And The End Of SaaS – But Is He Repeating Past Mistakes?
Posted on January 13, 2025
0
Here is an excerpt from a post I wrote in 2007:
“In my fall 2004 analyses of a study on the use of web-based applications I made the following observation:
It is my position that a true centralization of procurement objectives requires a decentralized architecture based on the real-world operating attributes of all transactional stakeholders, starting at the local or regional level. In other words, your organization gains control of its spend environment by relinquishing centralized functional control in favor of operational efficiencies originating on the front line. This is the cornerstone of agent-based modeling.
So while the ISM, CAPS and AT Kearney report is just now beginning to acknowledge the emergence of this dramatic shift, their ability to look beyond the conceptual level is significantly hindered by their respective technological blinders. This is reflected in their assertion that the next decade will be dominated by center-led organizations.”
In the follow-up post to the one above, I wrote the following:
“Throughout the research period (which was partly funded by the Government of Canada’s Scientific Research and Experimental Development – SR&ED program), we consistently looked for ways in which a buyer could reliably procure commodities on a real-time basis outside of the confines of a centrally negotiated contract. To do this effectively, the buyer must simultaneously engage critical stakeholders such as suppliers, courier companies and customs brokers, e.g., the Metaprise. Given that off-contract procurement was at epidemic levels (which negatively impacted both the buying organization and an increasingly skeptical supply base), it was essential to lessen rather than increase the purchasing cycle time. This cycle compression was a critical component in that we wanted to eliminate the compliance issues that had plagued so many initiatives (as it still does today). In August 2003, the new technology successfully went live in a production environment for the DND. In this test case, the public sector organization realized a year-over-year 23% cost of goods savings for seven consecutive years while simultaneously reducing the number of buyers required to manage the contract to 3 from an original 23. Delivery performance and product quality also improved dramatically.)”
Agentic AI And Agent-Based Models
In an October 2024 post, I wrote the following:
“Do you see the seemingly subtle yet critical differences between using a technology-led Agentic AI model and a human-led Agent-based model?
Suppose we as an industry don’t and continue to lead using an Agentic AI model. In that case, we will repeat the same pattern of technology-led implementation outcomes and have the same discussion regarding initiative failure rates 10 years from now as we are having today.”
Watch the following video from Microsoft CEO Satya Nadella within the context of everything above. My preliminary takeaway is simple: ERP, SaaS, AI, Generative AI, Agentic AI and anything else we develop tech-wise in the future outside of an Agent-Based model WILL ALWAYS FAIL!
I will use two words here to demonstrate why the CEO of a “technology company” may very well lead us further down the wrong road: Strand Commonality. Strand Commonality and its subsequent data output require the right “kind of human intervention” – otherwise, you will never get the correct data. Unfortunately, most humans are unable to effectively and consistently perform the kind of human intervention that is needed. I will delve more into this in future posts.
Microsoft CEO Satya Nadella Video (Dec. 2024)
30
Share this:
Related