AI — Almost Smart Enough to Replace Common Sense!

Almost!

If you’ve been following the doctor‘s deep dices in AI in Procurement () and AI in Sourcing () Today, Tomorrow, and the Day After over on Spend Matters Pro [membership required], you should be firmly aware that, in the true definition of the term, there is no AI in any enterprise software platform today. Period. In fact, what’s there isn’t even close.

You’ll also know we define AI today as assisted intelligence where AI helps you identify relevant information, trends, etc. much faster than you could on your own and augmented intelligence where AI supplements your intelligence with advanced algorithms and capabilities that allow you to make informed decisions many time faster than you could without the technology — and that even the most advanced platforms out there struggle to even deliver assisted intelligence. All the majority of platforms do is deliver RPA – robotic process automation, which is just a fancy way of saying they can automate complex workflows under a wide variety of conditions, provided somewhere along the line an expert has created enough rules and models for it to follow.

And if you’re willing to spend enough money, you too can buy a state of the art system that will tell you that people like to drink coffee in the morning and that if you have a lot of employees they will form a line to get it and they will waste a lot of time waiting for it if you don’t have a big enough coffee area and coffee pre-brewed and pre-poured. You know, the same thing any current or former Starbucks barrista would tell you for the price of a coffee.

And the doctor wishes he was joking. This was what he found in his twitter feed just before he had to stop everything and write this post!


Wow! How much did they spend on this software that documents basic process times and runs simple models to determine that some things take time and any action that reduces that time for multiple workers can save time and money? Hire any lean specialist to just walk around your operation for a day and you’ll get a hundred common sense recommendations like this.

the doctor thinks WeWork should say WeWereRippedOff!

Especially since the other “observation” was equally obvious and if they simply required people to book a room, log a count, and asked a secretary to review the logs for two weeks they’d learn in ten minutes that people prefer to work in smaller groups because smaller groups get things done!

I wonder if they are training these systems on Pigeon English? (At least we know the dangers that common from Pigeon droppings … )

Supply Market Intelligence … Harder than it Looks … But Possible with Modern Systems, Part II

See Part I for the story to date. Suffice to say that when the following are objectively analyzed, one can expect good market insights:

  • financial statements, particularly those from public companies (as false statements are a criminal offence for the CFO and CEO in some countries)
  • customer interviews, good or bad, as it’s a third party product/service view
  • performance reporting, as any hard metric is objective
  • internal stakeholder interviews, where the bias is minimized through targeted questions
  • price index data, that can be used to roll-your-own forecasts
  • public consumption data from government contracts, as they are great benchmarks

… provided one has the right platforms!

What are those platforms? Well, consider that the following sources are (primarily) numeric:

  • price index data
  • performance metrics
  • public price contracts

And the following sources are primarily (subjective) textual:

  • customer interviews
  • stakeholder interviews

And the following, final source is mixed:

  • financial statements

And that makes it pretty clear you need a platform that has the following if you want to process the price data:

  • A Great Open API
    as the price index data will be on multiple exchanges — which use different APIs, security protocols, currencies, and even data encoding formats and you will need to be able to easily retrieve and integrate all of it
  • Multi-Level Formula Based Cost Models
    to accurately capture and represent all of the commodity, component, product, and service costs that you need to track for cost estimation and analysis, bill of materials, sourcing, etc.
  • Powerful Analytics (Integration)
    you need to be able to store, analyze over time, and use multiple, multi-variate, statistical algorithms to detect trends and project them over time, as well as alter the assumptions, parameters, and model inflection points (due to predicted inflection events)

… and a platform that supports the following if you want to process the textual data:

  • advanced semantic processing
    that can extract key topics and opinions and classify them to process or technology, functional area, etc. (as well as identify incongruities)
  • advanced textual analytics
    the platform needs to be able to assign general descriptions numeric weights against important factors (perceived risk, customer service level, etc.) to determine if the general view is improving, weakening, or staying static
  • advanced sentiment analysis
    that can extract not only general opinions about a supplier, process, etc. but specific opinions about process, technology, etc. components — for example, the stakeholder might be soured on the relationship with a supplier because they have p!ss-p00r customer service but agree they make the highest quality parts (and would be usable if they ever bothered to answer the d@mn phone); just an overall negative sentiment of 0.6 is not that meaningful

… and to process financial statements, the platform needs to merge the advanced textual analytics to populate a standard financial model template, adding in any additional revenue or expense, asset or liability, etc. lines that are missing from the standard model so the books balance and can be analyzed.

So where do you find these capabilities today?

Well, as previously indicated, you will find:

  • advanced cost models in direct sourcing platforms that support full multi-level bill of materials
  • advanced forecasting in modern analytics platforms that support machine learning
  • advanced sourcing support given predictive costs in platforms that support strategic sourcing decision optimization
  • advanced document analysis in industry leading contract management solutions (which can be adapted to parse and analyze and break apart and score any document type given a template and samples)

In other words, modern Analytics, Optimization, and Contract Analytics solutions. And this is just another reason SI has been preaching advanced optimization and analytics since day 1.

Supply Market Intelligence … Harder than it Looks … But Possible with Modern Systems, Part I

Last year, about this time, we wrote a piece on Supply Market Intelligence and how it was Harder Than it Looks because there are a number of sources that might yield intelligence, including:

  • Suppliers,
  • Internal Sources, such as
    • internal stakeholders
    • performance reports
    • SRM programs
  • External Sources, such as
    • news feeds and alerts
    • price index forecasts
    • blogs and social media
    • peer companies
    • research services
    • advisory programs

… but not all are equal. And not all are fully accurate. For example:

  • Supplier company websites only show you what the supplier wants you to see, which is typically not the full picture, and maybe not even a true part of it
  • Internal Sources, such as
    • internal stakeholder interviews capture bias as well as expertise
    • performance reporting can only report on hard metrics the organization had the foresight to capture
    • SRM programs — and the insights yielded — vary by company and supplier
  • External Sources, such as
    • news feeds only cover the stories that interest the journalists
    • price index forecasts use in-house algorithms that are not disclosed that may not be accurate
    • blogs and social media cover the stories that can be sussed out by the bloggers and analysts

But some of them contain valuable data that when appropriately, and objectively analyzed, can yield good insights, as per our follow up post:

  • financial statements, particularly those from public companies (as false statements are a criminal offence for the CFO and CEO in some countries)
  • customer interviews, good or bad, as it’s a third party product/service view
  • performance reporting, as any hard metric is objective
  • internal stakeholder interviews, where the bias is minimized through targeted questions
  • price index data, that can be used to roll-your-own forecasts
  • public consumption data from government contracts, as they are great benchmarks

… and so on. But it can be pretty hard to make sense of all this … unless you have the right platform with the right capabilities. Now, it might not be a single platform from a single vendor and instead be a base Sourcing / Procurement platform augmented with multiple best of breed modules and API services from multiple vendors, and that’s fine. The point is that it’s possible to make sense of this with modern technology. What technology? How? That’s the subject of our next post.

Fifty Years Ago Today …

Soviet spacecraft Soyuz 4 and Soyuz 5 perform the first-ever docking of manned spacecraft in orbit and the first-ever transfer of crew from one space vehicle to another (which was also the only time such a transfer was accomplished with a space walk).

This was a historic event in space exploration as it is one of the capabilities necessary to have a(n International) Space Station and we’re not going to reach the age of extra-planetary supply management (Part II and Part III) unless we can build space stations and efficiently dock spacecraft with them on inter-planetary supply runs.

The Key to Successful Supply Management? No MoBAs, no PiMPs, no Paper Pushers, and no over-reliance on dumb bots.

It seems the list gets longer every year as those looking for a quick-fix try to take shortcuts to solving their problem that involve pushing those problems to third parties who are even less competent to solve them.

A few years ago we said the key for a successful supply management center of excellence was no M(o)BAs and no P(i)MPs!. This is because successful supply management relies on supply management expertise and experience, not on meaningless business models and knowledge-free project management frameworks. (Remember that SI still firmly believes that individuals that only have MBAs are just Master of Business Annihilation!)

This is because not only is it the case that you can’t manage what you can’t understand, but all you can do if you try is make it worse! Supply Managers are overworked and under-resourced, and any misstep has a ripple effect throughout the supply chain — one that can go from a minor delay to a major catastrophe. Management knowledge and project management skills are good things, but whereas supply chain is concerned, only if this knowledge and skill is added to a fundamental understanding of the supply management process that needs to be performed.

However, as we indicated last year in our post that The Key to Successful Supply Management? No MoBAs, no PiMPs, and no Paper Pushers!, simply eliminating the unknowledgeable MoBAs and PiMPs is not enough anymore. Paper pushers have to go to. There’s no time for tactical people who only receive, process, and send e-paper in a modern fast moving supply chain when the majority of this work can be automated by modern bots.

Today’s professionals need to be able to identify, implement, and make use of modern assisted and augmented intelligence solutions that can help them identify what needs to be analyzed, what needs to be addressed, what needs to be done, and the best ways to potentially go about it. The individuals who can do this are not PO paper pushers or AP invoice processors. They are knowledgeable and capable sourcing, procurement, and supply management experts who know their domain, and the tools, first and the business and project management second.

And they can’t be hampered by dumb bots. Dumb bots do poor invoice matching and create a lot of false positives to be unnecessarily checked. Dumb bots simply flag differentials between current and market price with no understanding of what the cause for the difference is and whether or not savings could actually be realized if a sourcing event was conducted. Dumb bots automate auction and RFX stages on a schedule, but don’t ensure that stages are complete or requirements are met. Dumb bots can extract potential terms, costs, etc. but make no sense of them and not even classify them properly. And so on.

Smart bots are needed, but dumb bots create more tactical work than they take away. So make sure they go with the paper pushers when you show them the door.