It used to be that data analysis that should be performed was avoided because it carried too much risk for the stakeholder due to the time and effort required.

Using original BI tools, when an analysis to determine a question, even a valid one, was likely going to take longer than X hours, and tie up limited resources on a quest that may or may not yield a return (even a decent five figure one), it had to be abandoned.

In addition, when an analysis to answer a question that could save six (or more) figures a year required a change to the organizational classification hierarchy to accomplish, which could require weeks to months to obtain (since classic BI systems could support one, and only one, schema that supported one, and only one, cube definition, and such a change would require the approval of all stakeholders), the analysis was abandoned because the effort to get the agreement could cost more than the analysis could return.

When analysis was not quick, easy, and always available, it didn’t get done when it should — and that’s any time someone had an idea that might save time, money, or both.

Today we have the same problem, but instead of a lack of analysis being performed, AI is performing too many and overloading the organization with “opportunities” of all shapes and sizes being pushed to the sourcing and procurement teams to pursue. And they are overloaded, with instructions to pursue and capture as many opportunities as possible.

Since most providers who provide AI Analytics provide hybrid solutions (where the Gen-AI is given access to traditional, deterministic, solutions that don’t make simple math mistakes), except when the LLM retrieves (or hallucinates) bad data or decides to skip using the deterministic solution for the calculation, most of the opportunities are real to some extent. And since most organizations don’t have good manual software, they’ll only have time to check a few, and when those come back good enough, the organization, out of time, will assume the rest are good (enough) as well and send them off to the sourcing and procurement team.

Some will pan out less than expected, some won’t, but since they generally won’t lose money (even though they won’t capture the savings they should), the organizational leaders will assume they’re doing well, that a manual analysis wouldn’t do much better, and that the volume of opportunities will make up for the lack of significant returns on the pursuit of any individual opportunity.

This is a problem for two reasons. First, the biggest opportunities will go undetected because the AI is running off of scripts, doing the same analysis over and over again, and without knowing all the variables, or having human intuition, won’t investigate where the real opportunities lie. It won’t see the downstream effects of a political tension that will result in a border closing, war, or strait closing that will jack some prices sky high unless demand is locked in now. Nor will it see the the impacts of too many data centres coming online at the same time as AI crash hits and not advise you to delay locking in long term data centre contracts until that happens.

Secondly, it will make mistakes, and sometimes the sourcing and procurement teams will spend a lot of time, possibly weeks or months, on exercises that result in new agreements that actually cost more money than the organization is paying (on average now) because they were undertaken at the wrong time, for the wrong demand, with the wrong suppliers … when existing contracts were still in place for partial demand that couldn’t be broken (without penalty).

These false opportunities won’t be caught because the manpower isn’t there to verify every opportunity (without the right software designed for rapid manual spend analysis), and, as a result, more manpower will be wasted verifying opportunities than just pursing what the AI spits out. However, the results won’t match the savings that would be achieved if the team only pursued verified opportunities (where you verified a real, significant, opportunity). But the lack of time and resources to verify (since management froze hiring to pay for the worthless AI) means the opportunities chased don’t get verified. When it takes more time to second guess the AI than to follow it blindly, the psychology of analysis is to not do it, just like 20 years ago the psychology of analysis was not to do it.

In our last post on how AI increases the need for manual spend analysis, we noted that the claims that AI negates the need for manual spend analysis is all lies, damn lies, and statistics (of the worst kind).

AI throws more “opportunities” at you than you can process with spreadsheets and traditional AI tools, and it does so faster than an entire team of analysts working in the basement, and, unlike the opportunities coming from the analysts which are at least plausible, the “opportunities” coming from the AI may be completely hallucinatory and a complete waste of time (compared to the opportunities from the junior analysts that may not be worth the effort but are actually real).

So you need to be able to investigate and verify them very, very quickly — which means you need a tool that can:

• rapidly suck in a dataset, or multiple
• allow for rapid (re)-classification and (re)-categorization on the fly, using both pre-defined and on-the-fly rules, as well as Gen-AI for suggestions (with quick acceptance or rejection)
• allow for the quick construction of cubes (and sub-cubes) and views (and sub-views) to verify spend, variances, and potential opportunities against market prices
• allow for easy generation of derived dimensions and derived cubes with what-if scenarios
• allow for easy visualization and cross-comparison of scenarios to identify and confirm true opportunities

And then when the opportunities don’t pan out, you need to be able to quickly

• replicate the data set and rapidly (re)-classify and (re)-categorize on the fly, using both pre-defined and new on-the-fly rules
• create new cubes (and sub-cubes), views and sub-views on the recategorized data
• create new derived dimensions (and derived sub-cubes) based on what-if scenarios
• cross-compare and visualize the new scenarios to identify real opportunities

Or you will never realize any value from spend analysis, especially if you are using AI. The only difference now vs. 20 years ago when you desperately needed a good spend analysis tool was that 20 years ago you didn’t have any opportunities unless you had the time to look for them. Now, you don’t have any opportunities unless you have the time to dig through all of the fake opportunities generated by AI to weed out the real ones from the hallucinations. The value curve is the same, only the ends have switched positions.

Everyone with an AI tool is proclaiming how their tool negates the need for manual spend analysis and, beyond that, how it negates the need for sourcing (because their Agentic solution can create the event and execute it automatically), and how it can even handle the contract generation, negotiations, and signing. How it will save you so much time and money that you will never need to do manual spend analysis again.

And it’s all lies, damn lies, and statistics — of the worst kind — since it usually involves probabilistic Gen-AI that often hallucinates market conditions, best practice, and true innovation.

It’s true that an agentic solution can run pre-packaged spend analysis against all of your spend once it has been classified, identify the products and categories with the greatest variance, compare the average price paid to current market prices, suggest opportunities, and push the products into the sourcing tool (and invoke the sourcing agent).

But here are the problems:

• even assuming it’s ranking opportunities by the savings today, that doesn’t mean it’s the greatest savings tomorrow; there could be contracts in place, declining demand, or rising inflation in the category
• similarly, products or categories passed over could be the greatest opportunities due to increasing demand, lack of long term contracts, or stagnation in pricing
• without verification, the “market prices” could be quite off actuals as it could be “public” pricing only and private is quite cheaper, especially with volume discounts
• the opportunities are not real unless they are with suppliers you can actually source from (and auto-identification is not actual product verification)
• and just getting auto-responses to auto-requests doesn’t guarantee you get the responses from the suppliers with the best products and prices

And that’s just the beginning.

Here are the missing pieces.

• it’s working on a predefined categorization (yours, the vendor’s, or, even worse, a random AI one); sometimes the best opportunities come from a categorical reorganization that allows you to bundle the right mix of products for the suppliers you’re inviting
• suppliers don’t auto-bid in response to auto-RFI requests, vendors do — and for strategic categories, you need strategic suppliers
• you can use agentic CLM to assemble clauses or send and retrieve documents and even do an initial analysis, but since the review is based on Gen-AI LLMs that can hallucinate as often as not, a review can be helpful, but it’s not a real review
• the data ingested for product matchings may or may not be accurate, especially since suppliers or distributors can put want they want on their site, or, more importantly, leave out key details

It’s great to use AI to find potential opportunities, but you can’t use AI to verify, and you definitely can’t use AI to capture them. Manual spend analysis has just become more important than ever, as has a tool that allows you to do it as efficiently as AI appears to. Fortunately, a such tools do exist — although you may have to work hard to find one.

In this age of AI hype, every vendor with an “Agentic AI” or an “AI Employee” is claiming their LLM-based garbage app is going to provide you with the “spend intelligence” you need for your Procurement success. The reality is that it is as much hogwash today as it was 20 years ago when I wrote that There’s No Such Thing as Spend Intelligence.

The words I wrote exactly 20 years ago today still stand.

There is no such thing as a spend intelligence solution.

Let’s start with the definition of intelligence.

Intelligence is a most complex practical property of mind, integrating numerous mental abilities, such as the capacities to reason, plan, solve problems, think abstractly, comprehend ideas and language, and learn.

And since spend management solutions are software, let’s review a definition for software.

Software is the (collection of) program(s) that enable a computer to perform a specific task, as opposed to the physical components of the system (hardware), where a program is the collection of source code and libraries which have been compiled into an executable or otherwise interpreted to “run” in (active) computer memory, where it can perform both automatic and interactive tasks with data.

Simply put, intelligence is a property of mind and software is a property of machine. And despite the efforts of the artificial intelligence community, I do not expect the property to cross the chasm anytime soon. Artificial intelligence is simply a collection of very sophisticated algorithms processing large data stores, instruction sets, and probabilities very quickly to come up with reasonable responses to queries – it is not thought, although it might appear to be thought since today’s computers can perform billions of calculations in a second.

The only difference is that today’s processors easily do hundreds of billions of instructions per second and with multi-threading and massive parallelization, average racks do hundreds of trillions of calculations per second and custom AI Data Center racks do quadrillions of calculations per second. But calculations do not equal intelligence. And since the models are still probabilistic, they haven’t gotten any better and, in fact, with LLMs, they’ve gotten worse!

In fact, with hallucinations a core function, not only can you not depend on the math (which these models get wrong all the time) but you can’t even be sure they’re working on the right data. We’ve went from insight to inaccuracy. That’s not intelligence. That’s idiocy … at it’s finest!