Category Archives: AI

AI will be Procurement’s Downfall … And There May Be No One to Save It If That Happens

A post this summer by David Elkington on LinkedIn on the impacts of AI on an Ivy League class is truly terrifying. It’s not just overworked and lazy adults turning their lives over to hallucinatory chat, j’ai pétés and clods, but our next generation as well.

In his post, David Elkington summarized how Brown University professor Roberto Serrano has proved it. When, after 20 years of teaching a course, he offered his first take-home midterm, enrollment tripled and the average shot up almost 40% to 96 (A++). He then had his Teaching Assistants run the exam through ChatGPT, from various accounts, and they were able to reproduce the majority of proofs that dozens of students submitted.

So then the professor announced he would be returning to tradition, with an in-person final. Twenty percent of enrolled students dropped, and ten percent didn’t show up. The final average was 48.6, an F and a drop by half.

The majority of students had obviously not only outsourced their studying, but their thinking as well. After all, the course was “Welfare Ecomomics”. Formally defined as using microeconomic techniques to evaluate societal well-being and guide public policy, and relying on methods like Pareto efficiency to help analyze trade-offs, neither the math nor the analysis is particularly hard compared to other areas of economics, undergraduate mathematics, or even philosophy. The trick is a bit of study and a lot of logical, deep thought to think through and work out the consequences and likely reactions to a decision. Especially when there aren’t always absolutes like there are in physics and chemistry. When the answer is human intelligence, AI fails.

And that failure will carry forward into the real world.

Everyone forgets that computers were designed to compute — and do so at a speed and certainty that we cannot do as humans. That’s what they’re good at. We’ve spent over a trillion hours as a species researching computing science, writing, and re-writing code and not only improving the capability of computers but pushing the limits of what computation can do.

The first few decades were spent perfecting the hardware, the error correcting code, the operating systems, the core compilers, and testing methodology so that the chances of a well defined program on a stable operating system on well made hardware making a calculation error is now one in one quadrillion operations. Then decades were spent taking advantage of the fact that computers can repeat the same operation that many times flawlessly where as we, as humans, struggle to do the same calculation ten (10) times in a row without error. And we created not only business software to make our business operations more efficient, but communications software to make our personal lives more connected and easier, and education software to help us get smarter.

But then someone decided that if we created a big enough next-generation neural net and fed it enough data it would magically become intelligent, unleashed it on the world, and when it, for the first time since the concept of AI was conceived almost seven decades ago, natural language was produced that was indistinguishable from an average person (whose IQ was already deteorioating thanks to Twitter), someone else, of very limited intelligence, thought it was intelligent. Then the hype-mongers decided to proclaim this story as gospel, and convince the masses it was the word. And those who believe it will lead us to our collective downfall.

Only Human Intelligence (HI) will allow Procurement to deal with the constant turmoil of today’s world and adapt to the changing reality. Without it, they will not be able to secure supply at a cost the organization can afford, which is not only the core function of Procurement, but fundamentally the core function of a business. (If an organization cannot secure supply better than an individual, there is no reason for the business to exist.)

AI will be Procurement’s Downfall, and to be honest, at the rate things are going, Humanity’s downfall as well.

Speed is Not Strategy!

Everyone is still jumping on the I2O (Intake-to-Orchestration), Agentic, and AI-hype bandwagons (and when one heads for the cliff, jumping on the next one) because they see the speed boost it is supposedly giving their peers and the purveyors of this depraved mind porn are promising that it will take them from being tactically-oriented fire-fighters to strategic spend sleuths who will magically save money while reducing stock-outs, risk, breakage and spoilage, service errors, etc. In other words, simply by adopting the tech-du-jour, and getting faster, they will magically improve across the board.

Anyone who isn’t suffering severe (Gen) AI-induced cognitive atrophy should realize this is the biggest crock of bullshit we’ve yet been served in the Procurement space.

It doesn’t matter how fast you go, how many RFPs you can automate, how many suppliers you can search, and how many risks you can theoretically track, it matters how many truly strategic RFPs you can successfully execute properly, how many appropriate suppliers you can develop relationships with, and how many risks you can mitigate. Guess what? Strategic RFPS can’t be successfully executed with speed, only finesse. Relationships take time, as they take trust, and that takes time. Mitigation plans take deep thought, scenario planning, and even war gaming — to understand what will happens when man-made wars or natural disaster (where the environment itself declares war on your supply chain) disrupt your supply chain. This mitigation planning takes deep thought and time to work through the details — time that cannot be reduced by AI. AI can only regurgitate previous responses to previous situations that worked, it cannot predict new eventualities or determine what would work best against them because that requires real human intelligence (HI!).

What everyone forgets is what inevitably happens when you run too fast — you get exhausted too soon, and collapse. There’s more than one reason the tortoise will beat the hare every time. It’s not just because the hare is cocky, it’s because the hare will try to rush that which cannot be. There’s a deeper meaning to Aesop’s Fable than you are taught in grade school.

It’s not how many RFPs you can automate, it’s how many you run where you get real results. You get real results only where the category is strategic, the spend significant, the supply chain unoptimized, and the opportunity for improvement real. You don’t get results simply sending every category to an RFQ and shaving a percent or two off the price. You get results by improving quality, supply assurance, price predictability, risk reduction, and real human relationships. This requires an investment in time, not tech.

Speed does not help, and all technology focussed on simply increasing speed does is hinder.

“The Suite Was Supposed to Settle Everything.”

This is a requested response to Paul Marty’s inquiry: Is Source-to-Pay Dying.

We’ve heard the “one app to rule them all myth” since the late 60s when IBM said the MRP would control the operations and no other app would be needed. This was rekindled in the mid 80s when Oracle said the ERP would control the organization and no other app would be needed. Then in the late 90s when the web would provide a common interface to everything. Then the late noughts / early tens when the Suite would solve everything. It’s the myth that keeps on giving.

The reality is that any new monolith application model only solves all the problems that came before, not the problems you have now, and not the problems that are coming.

That doesn’t mean suites don’t have value, or that suites won’t provide more value than any other alternative, just that it’s not a panacea. Determining whether it provides more or less value requires understanding the true total value realized (taking all costs and benefits into account) compared to the true total value of the alternatives, many of which are unproven.

“Is The Economic Value Moving Somewhere Else?”

First off, let us note that, fundamentally, the new AI orchestration craze is no different, as it ultimately depends on a protocol that can bind everything together. This isn’t new either. Remember CORBA? Probably not, as most of you probably don’t even know what it is. Look it up. At the dawn of the dot com era, it was going to solve everything. It made all the same promises MCP is making now.

The reality is that the fundamental truths of technology have not changed.

> If it sounds too good to be true, it is.

> Technology can’t advance any faster than discovery (either mathematical computing theory or practical scientific discoveries that allow for new engineering that can build better machines to support better models), and that has, and will, always be slow. How many hundreds of thousands (or millions of years) has it taken us to compute a theoretical warp bubble, and how many decades/centuries more before we can achieve it? (We’ve went from conjecture with the introduction of Star Trek 60 years ago to theoretical possibility 5 years ago … but now we need to create the energy source, the materials that could withstand it, etc. etc. etc.)

> … businesses not only need to make money every year, but are punished by the capitalistic economy if they don’t; so even if all they have is the same old sh!t, they still need to sell it, and the best way to do that is to slap a new coat of paint on the old Mustang and tell you it’s a brand new model (even though the engine hasn’t changed in almost a decade)

This doesn’t mean that AI orchestration doesn’t provide value, just that it’s not the ultimate answer and that you have to truly understand it’s true cost vs its true benefits and, where autonomy is granted, true risks before you can make a decision.

“What if the employee does not need to enter the procurement system at all?”

That doesn’t change the fact you still need a procurement system.

Just because an intake layer can work across applications, and just because an AI agent can complete a routine workflow without ever displaying the myriad of screens suite vendors spent years (if not decades) refining, that doesn’t mean that the core functions go away. Either those have to be part of the intake/orchestration system (in which case it’s really a suite with a new kind of interface) or you still need the underlying best-of-breed / suite systems, which are only accessed through their APIs.

In the latter case, the true economic value remains in the best-of-breed/core systems … and the only value of the intake/orchestration/AI agent is the cost of user time reduction (i.e. 5 FT AP agents to manually process all the invoices vs 1 FT AP agent to manually deal with the exceptions). So if the cost of the software is greater than the manpower saved, it’s negative value.

This assumes that the agent is deterministic and will do all processing according to user defined rules and in user defined bounds. If you are using LLM-powered agents and giving them autonomy, there is no guarantee they will follow all rules and stay without bounds — autonomy means that they can decide only 10K over their approval limit is ok because it appears to be a known supplier, onboard an uncertified supplier, issue a new PO to a supplier who is late on other critical deliveries, etc. which can result in significant losses.

Which brings us to the most important question:

“ONE INTERFACE > MANY SYSTEMS > WHO DECIDES?”

This is a key question for multiple reasons:

1) You can’t cost / benefit a system without knowing.

As above, if the decision is ultimately human, you can accurately cost the system and determine it’s fixed (workforce reduction) benefit as well as it’s expected benefits from (weighted) average results and track true benefits over time.

But if the decision is ultimately the system, you can’t truly estimate the benefits as you don’t know what it will do, how many mistakes it will make, and what the cost to clean them up will be — and if it uses LLM technology, you can’t even cost the system as you don’t know how the token/compute cost is going to increase over time as you don’t know the token growth that will be required or how soon you’re going to have to pay true compute costs (the big AI companies can’t keep losing $4 for every $1 of revenue, and with today’s energy and water costs, the true compute cost is about 5X what most of the players are charging … and this will only increase over time.)

2) You can’t quality control without absolutely knowing who decides what, what the range of options are, where, and how to encode the checks.

Which means that:

“THE CENTRAL QUESTION … IS WHO OWNS THE REASONING THAT DETERMINES WHAT PROCUREMENT SHOULD DO”

Especially when the answer to:

“Are Twenty Agents Really Twenty Opinions?”

is NO!

I know Jon W. Hansen might disagree with me here, since he uses multiple systems to arrive at his (starting) algorithmic analysis of the marketplace regularly, but agents reason off of the data they are given.

When you interconnect them, they share “data” and “reasoning” continuously. Also, even though they are ultimately sophisticated probabilistic models no human can fully understand, the core is similar to deep multi-level feedback neural networks. This means that common “data” and “reasoning” increases the weighting on that “data” and “reasoning”, increases similar output, which increases similar input to the minority models, which causes that “data” and “reasoning” to become more heavily weighted in them. Or, in simple terms, a majority of agents will bully the minority of agents into their viewpoint until there is one opinion, not twenty.

Even if all 20 agents start out different, as they share “data” and “reasoning”, there will be some commonality, and as the “data” and “reasoning” flows back and forth, the weighting of the commonality will increase until it snowballs into a majority, given enough exchanges (over time).

While some “experts” will try to tell you this all but certainly guarantees you will get the best results once the agents agree, that is ONLY true if the common “data” and “reasoning” is accurate. Some agents will have bad / fake / hallucinated (wrong) data and others will share faulty “reasoning” because they don’t actually reason. (They predict arguments … think of them as a super powerful fancy version of next word suggestion on your iPhone, except they are predicting answers to questions and sometimes even the steps involved to get there!)

While individual agents sometimes work well if they have enough raw question/answer input/output previous step/next step data top work from (and that’s how they create mathematical “proofs” — they can scan millions of research papers and proofs, build up all the associations as input/output pairs, and then see if there is a way to chain them all together from a question to an answer where the output of one step in one paper is the required input of a step in a paper in another paper, possibly published in an unrelated journal in an unrelated discipline), they sometimes don’t work very well at all!

They’ll tell you to eat one rock a day. Or that strawberry only has 2 rs. Or, because they don’t understand reality, when asked, they will tell you that “the woodchuck could carry 57,600 pounds of wood in a day” when asked “How much wood could a woodchuck chuck if the woodchuck was 10 lbs?” (I didn’t make this up, DeepSeek did!) (And, even worse, it gave some extremely ridiculous “reasoning” to back it up which might fool someone who didn’t think twice because the logic was pretty good, while the data it worked on was ridiculous [and obviously hallucinated]. I should also point out that this was a result “I” totally expected because there shouldn’t be much “data” on “how much wood a wouldchuck could chuck if a woodchuck could chuck wood” on the internet [which is it’s training data], because we all know a woodchuck can’t chuck wood but the AI doesn’t know that.)

Thus, whether you use one agent or twenty, you will get one opinion, and one result, which could be a total hallucination if none of the systems had enough correct “data” and examples to “reason” on.

Given that, as Paul Martyn points out, the best decisions are made when there is

  1. A diversity of experiences and information to draw on
  2. Independent thought by the participants (that can provide truly different views to the ultimate decision owner)
  3. Decentralization of knowledge for risk mitigation

and that today’s AI, which is NOT intelligent, cannot be trusted to always make the best decision as

  1. all agent systems have access to the same data on your network and the web
  2. agents cannot think, only produce “outputs” based on different weightings of “inputs”
  3. agents with minority data/reasoning can, and will, have at least their reasoning, if not their data, overwritten by the majority

This brings us to the following conclusions:

1. Traditional Suite and BoB solutions ARE still needed, even with orchestration and agents (especially since the later should be restricted to either fully deterministic processes or processes where the cost to fix/clean up after an occasional bad decision by software is less than the cost of having a team of people do that process manually)

2. The New Hotness Hype will always take sales away from established players, as the lazy and greedy look for that panacea solution that does not, and cannot exist … but when those solutions fail to achieve the result, some of those sales will return (as the only other option will be to build the missing parts into the acquired system, and why waste money trying to extend a system that’s fundamentally broken?)

3. “The Market is Not Dying”, as Paul Martyn states, but it is shifting. Suites will become the transaction engines and systems of record in organizations that experiment with new interfaces, orchestration, and AI agents. Hybrid suites, which embed orchestration and intake, will take over in organizations that want modern connectivity and interface but want to limit the number of vendors they need to deal with. And traditional suites will gain traction in highly regulated industries and governments where everything has to be 100% predictable and in-line with thousands (if not millions) of pages of policy regulations!

As Jason Busch likes to say, “many things can be true at once” and as long as those things are not mutually exclusive, they will be in the market.

(P.S. Thanks for taking up my day with no warning, Paul!)

If You Can’t Do The Job Without Tech …

Then you can’t do the job with tech.

I’ve said it before and I’ll say it again. If you can’t do the job without tech, then you can’t do the job with tech.

All tech does is automate tasks and workflow processes and allow you to speed them up by a factor of 10 to 1,000,000 plus (depending on the tasks and workflow processes).

It doesn’t make tasks or processes better or worse (unless you Gen-AI, which typically makes it worse) — just faster (unless, of course, you change the process when you implement the tech and make the task or process better or worse).

Now, you’re probably asking why it’s not enough to design the process and simply install the tech. That’s because there’s a difference between designing a process and executing a process. Anyone can design a process at a high level given a highly set of requirements. Not anyone can execute. Among those, even less can execute it effectively.

Only those who can execute it effectively are fit to do the job, regardless of what tech the organization has or does not have. That’s because, if you don’t know how to do the job without tech (and certainly without AI), you don’t understand what the job really is. Business was conducted for thousands of years without modern tech, and computers only became generally available in big business in the 80s and mid-size businesses in the 90s and small businesses in the 2000s. That means everything you do today with modern tech was once done without modern tech (and, 150 years ago, without any tech that we would consider modern at all).

That understanding means you understand the core of the function. You know what has to be done, how, where the time suck is, what can be automated, how, and why. You also know how to verify correct automations has been implemented, and where the results require human interpretation, review, and decisions — and, if necessary, make those interpretations, review, and decisions.

Without that understanding, you can’t properly make use of tech and you definitely can’t make use of AI. Sure some tasks might be 10 to 1,000,000 times slower (and sourcing optimization will be out of the question, just a few bid comparisons), and you shouldn’t do them by hand unless necessary, but the point is you need to be able to — otherwise, you can never judge what the system does. Iin the age of hyper-fast LLM hallucinations, only good human decisions will allow your organization to succeed.

The Procurement Knowledge Devolution

The internet was supposed to kick off the procurement knowledge revolution, allowing Procurement pros to quickly:

  • find out about best practices
  • get commodity and market pricing for products and services
  • build reasonable (should) cost models
  • find out about new suppliers, carriers, and consulting partners
  • research new products and services
  • hold truly global sourcing and procurement events in real time
  • effectively communicate with, manage, and develop suppliers
  • etc.

For twenty years, that’s what it was. Procurement departments who learned how to use the internet properly for research, deployed the right SaaS to support their processes, and identified the right data for their processes were successful in their endeavours.

But then came Gen-AI LLMs, chatbots were replaced with chat, j’ai pété‘s and clod‘s, and knowledge was replaced with whatever content the LLM generated. Maybe it was correct, maybe it was mostly correct, and maybe it was a 100% fabrication — a hallucination if you please. The problem with LLMs is that they are NOT intelligent. They are essentially super sophisticated multi-level cross-connected deep neural nets that go beyond classification to generation of responses built up from sub-responses built up from deep training on incredibly large data sets.

Therein lies all of the problems. It generates built on random probabilities. Those are dependent on what’s in the training data set, what questions were asked, what results are reinforced, and how it’s used. If the training data is bad and full of bad data and falsehoods, the chances of incorrect, and even dangerous, responses being generated are quite high. If the training was biased, the output is likely to be very biased. And if it’s not “trained to please”, the models are fundamentally designed to “learn to please”, so if computations that are a complete lie will increase utilization of, and faith in the model, that’s what will happen.

Moreover, every vendor is now believing the hype from the big LLM players, treating the technology as real Artificial Intelligence (when it should be called Artificial Idiocy), and trying to plug it in everywhere … promising that it will provide their clients with true natural language interfaces, agentic tech, and even BS AI Employees. Those who are adopting it are literally getting dumber by the day. Not only has the cognitive impairment, atrophy, and potential long-term decline from regular use been well documented, but the failure rate has been well documented as well with MIT and McKinsey studies demonstrating success rates of 5% and 6% successfully. Most pilots are being abandoned, sometimes before they even begin, because the tech isn’t even good enough to put into employee’s hands for the tasks the overpriced Big X consultancies claimed the LLMs would be perfect for.

Furthermore, even when organizations are smart enough to ignore Gen-AI, over-automating using the most advanced last-gen (A)RPA and AI technologies will still give Procurement teams a false sense of security and, due to their very low error rate, as time goes on, the team’s skills will go rusty and their ability to deal with true exceptions quickly disappear, especially as the old Pros (get forced to early) retire and the younglings have never dealt with exceptional situations.

The age of AI hype has ushered in a knowledge devolution faster than any age that has come before.

I hope the profession can survive it!