Latest / Speaking of Supply Chain / The Real Way AI Is Changing Supply Chains

Transcript

The Real Way AI Is Changing Supply Chains · Speaking of Supply Chain

• 0:00
  Speaker 1: Suddenly the plan manager went from why did this happen?
• 0:03
  Speaker 1: To we saw this coming and here's how we are
• 0:06
  Speaker 1: addressing it. So AI acts as sort of a therapist
• 0:10
  Speaker 1: for many of the planners, saying, hey, everything's fine, We're
• 0:14
  Speaker 1: going to go through this together and here's how we're
• 0:17
  Speaker 1: going to fix it.
• 0:19
  Speaker 2: Hello, and welcome to Speaking of Supply Chain, where meboch
• 0:22
  Speaker 2: experts from around the world unpack the real stories, strategies, insights,
• 0:27
  Speaker 2: and innovations shaping the future of supply chain. I'm your host,
• 0:31
  Speaker 2: Ellen Wood. Supply chains generate massive amounts of data, but
• 0:36
  Speaker 2: all too often that data fails to turn into real decisions.
• 0:41
  Speaker 2: As AI tools mature and analytics platforms evolve, the challenge
• 0:45
  Speaker 2: isn't collecting the information, it's knowing how and when to
• 0:49
  Speaker 2: act on it. In this episode, we're going to explore
• 0:52
  Speaker 2: how to close the insight to action gap and where
• 0:56
  Speaker 2: AI is truly transforming supply chain planning and perform romans.
• 1:01
  Speaker 2: Joining me for today's discussion is my colleague beckball at Ebdiknov.
• 1:06
  Speaker 2: He is the generative AI Solutions lead for the US Office.
• 1:10
  Speaker 1: Welcome back, Thank you, Ellen, excited to be here.
• 1:14
  Speaker 2: It's great to have you, so tell us a little
• 1:17
  Speaker 2: bit more about your role in Meebok and what we're
• 1:22
  Speaker 2: doing with AI.
• 1:24
  Speaker 1: My path into supply chain at Kneebok and AI in general,
• 1:30
  Speaker 1: actually started with something simple, noticing how much time organizations
• 1:35
  Speaker 1: spend just figuring out what happened yesterday in their warehousing solutions.
• 1:40
  Speaker 1: It's an everyday problem, that every day challenge that happens
• 1:44
  Speaker 1: to factor into logistics. So I've always wanted to work
• 1:49
  Speaker 1: at a company that was future forward thinking, and Kneebok
• 1:54
  Speaker 1: was just the right place where I could practice my
• 1:58
  Speaker 1: expertise and bring the value add to more services that
• 2:04
  Speaker 1: we have here.
• 2:05
  Speaker 2: Awesome, So let's get into it. Despite the explosion of
• 2:11
  Speaker 2: data and all of these digital tools that are now
• 2:14
  Speaker 2: available on the marketplace, many supply chain leaders struggle to
• 2:18
  Speaker 2: act on any of the insights that they're getting from
• 2:22
  Speaker 2: these tools, from this data. Why is that great question?
• 2:27
  Speaker 1: I would say there are three big reasons why this
• 2:31
  Speaker 1: gap still exists. Number one, I would say companies don't
• 2:35
  Speaker 1: have a data problem per se, they have a fragmentation problem.
• 2:40
  Speaker 1: So the data is spread across different systems like ERPWMS,
• 2:46
  Speaker 1: planning tools, Excel spreadsheets that we all love I would say,
• 2:52
  Speaker 1: mis on the shop floor, and none of them speak
• 2:55
  Speaker 1: the same language. So a planner might get a forecast
• 2:59
  Speaker 1: from demand planning, but labor schedulers live somewhere else, skew
• 3:05
  Speaker 1: profiles live somewhere else. Again, so by the time all
• 3:08
  Speaker 1: of this get reconciled manually, the insider is already outdated.
• 3:13
  Speaker 1: So that's number one. I would say Number two is
• 3:17
  Speaker 1: dashboards does not equal to decisions, meaning most organizations think
• 3:24
  Speaker 1: they just need more dashboards. The more dashboards they will get,
• 3:28
  Speaker 1: the more they can capture this ever ever infinite amount
• 3:33
  Speaker 1: of data that they will do something. But but they
• 3:36
  Speaker 1: don't realize that dashboards just tell them what happened, not
• 3:40
  Speaker 1: what to do next. So of course that can change
• 3:44
  Speaker 1: with the size of the organization. So the bigger the organization,
• 3:48
  Speaker 1: the more dashboards you will obviously need. But at some
• 3:51
  Speaker 1: point you'll have to stop and ask yourself, is this
• 3:54
  Speaker 1: enough to make a decision or do I have enough
• 3:58
  Speaker 1: metrics to base my decisi So I would say that's
• 4:01
  Speaker 1: number two. Number three, I'll say this decision ownership is
• 4:07
  Speaker 1: very unclear. Most of the time, organizations know that even
• 4:12
  Speaker 1: if inside exists, no one knows who should that act
• 4:16
  Speaker 1: and how to act. So that would be my three
• 4:20
  Speaker 1: reasons why that gap still exists to action.
• 4:25
  Speaker 2: So it sounds like one of the ways that AI
• 4:27
  Speaker 2: can help close that gap is by making all of
• 4:32
  Speaker 2: the language or all of the data from these disparate
• 4:35
  Speaker 2: systems talk about the same thing and organize it into
• 4:40
  Speaker 2: a way where it's at least cohesive across the organization.
• 4:46
  Speaker 2: Is that is that accurate or is that still going
• 4:51
  Speaker 2: to be a gain? Is there's still going to be
• 4:54
  Speaker 2: a gap there just because of the different kinds of
• 4:57
  Speaker 2: data that's there.
• 4:58
  Speaker 1: You bring up a good point. Actually, this is very important.
• 5:03
  Speaker 1: As most of us know, whenever we put something into AI,
• 5:08
  Speaker 1: the quality of our inputs matters. So this is a
• 5:14
  Speaker 1: story that brings into one of my experiences. Whenever I
• 5:20
  Speaker 1: walk into the warehouse for the first time, I can
• 5:22
  Speaker 1: always always predict the first conversation. Somebody pulls up the
• 5:26
  Speaker 1: dashboard and says, we knew the bottleneck was coming. We
• 5:31
  Speaker 1: didn't know how bad it would get or what decision
• 5:35
  Speaker 1: to make about it. So there were multiple systems that
• 5:39
  Speaker 1: they were pulling it from. It's not that the data
• 5:42
  Speaker 1: wasn't necessarily there, it's that no one connected the dots
• 5:46
  Speaker 1: quickly enough to make that call. Hey, this system is
• 5:51
  Speaker 1: not talking to this system.
• 5:53
  Speaker 2: So even if we harmonized all of that information, it
• 5:58
  Speaker 2: still got to be there in a timely man. They've
• 6:00
  Speaker 2: still got to pull it at the right time. They've
• 6:02
  Speaker 2: still got to you know, make it make sense, and
• 6:04
  Speaker 2: then to your point, have someone make a decision about
• 6:07
  Speaker 2: it exactly.
• 6:09
  Speaker 1: I would say this is the biggest point.
• 6:12
  Speaker 2: So there's so much hype around AI, and it's great
• 6:17
  Speaker 2: in terms of my role marketing, the large language models,
• 6:22
  Speaker 2: the chat gbts. There's clearly value there for these types
• 6:28
  Speaker 2: of roles. But when it comes to supply chain, and
• 6:32
  Speaker 2: again to your point, the decision making of which everyone
• 6:35
  Speaker 2: seems to be hesitant to actually make decisions, is it
• 6:40
  Speaker 2: already creating value in supply chain?
• 6:43
  Speaker 1: Actually a lot of value. A lot of the value
• 6:47
  Speaker 1: comes in warehousing and manufacturing logistics environments where there's so
• 6:52
  Speaker 1: much human involvement. There is an opportunity for AI to
• 6:55
  Speaker 1: step in and add that real automation, real game where
• 7:02
  Speaker 1: we can see something that would take the human many
• 7:06
  Speaker 1: hours to do. So let me give you an example.
• 7:10
  Speaker 1: One of our past projects involved a distribution center that
• 7:15
  Speaker 1: we supported during the struggling times. They struggled with labor
• 7:20
  Speaker 1: swings because volume peaks we weren't predictable at all. So
• 7:25
  Speaker 1: humans were forecasting labor weekly, but inbound variability made that impossible.
• 7:32
  Speaker 1: So for them, AI changed the game. It started forecasting
• 7:37
  Speaker 1: hourly throughput based on their SKW mix, on their order profiles,
• 7:42
  Speaker 1: on their seasonality, receiving patterns past performance. So the supervisors
• 7:48
  Speaker 1: in the end, after we have finished our project, went
• 7:52
  Speaker 1: from reactive firefighting to actual proactive orchestration. So we have
• 8:00
  Speaker 1: made more predictable operations and up to twenty percent reduction
• 8:06
  Speaker 1: in overtime.
• 8:07
  Speaker 2: That's fantastic. So we've already moved that needle from reactive
• 8:13
  Speaker 2: to at least predictive. What about prescriptive where it's telling
• 8:19
  Speaker 2: you what you should be doing in the future based
• 8:22
  Speaker 2: on not only what do you need to do now,
• 8:25
  Speaker 2: but what should be changing any value there yet.
• 8:29
  Speaker 1: That's a great point. The prescriptive inventor optimization is one
• 8:33
  Speaker 1: of the past projects, so that would mention that deserves
• 8:38
  Speaker 1: to be mentioned here. One client hand constant stockouts on
• 8:43
  Speaker 1: a set of critical SKUs, not because demand was really
• 8:46
  Speaker 1: crazy or anything, but because buffers were set manually months ago.
• 8:51
  Speaker 1: So we used AI to analyze their consumption variability, their
• 8:57
  Speaker 1: production cadence, and lead times and began commendtic dynamic inventory adjustments.
• 9:02
  Speaker 1: So this took them from reactive for plenishment to actual
• 9:07
  Speaker 1: predictive optimization, which is exactly where the real world AI
• 9:12
  Speaker 1: value comes in.
• 9:13
  Speaker 2: That's awesome. So from your vantage point working with clients,
• 9:20
  Speaker 2: what does an organization need in terms of data maturity
• 9:24
  Speaker 2: and organizational readiness? Because it is it isn't just about
• 9:28
  Speaker 2: the data obviously, you know, if they've got some of
• 9:30
  Speaker 2: these systems, the data exists, but the organizational readiness and
• 9:37
  Speaker 2: let's say, the people readiness to use these AI capabilities
• 9:46
  Speaker 2: within these supply chain functions in order to truly benefit
• 9:51
  Speaker 2: from it. So it's it's not just about you know, hey,
• 9:54
  Speaker 2: it's possible, it's you know, are they are they ready?
• 9:59
  Speaker 2: Are the ready to take on this change? Because it
• 10:02
  Speaker 2: is a huge change to start looking at data from
• 10:06
  Speaker 2: this perspective, I should say, and to your point, trusting
• 10:12
  Speaker 2: the data enough to do something about it, I.
• 10:15
  Speaker 1: Would totally agree with you. Having organizations be ready for
• 10:20
  Speaker 1: the AI revolution is one of the equations to make
• 10:23
  Speaker 1: it a successful adoption. So I would say it's a
• 10:27
  Speaker 1: blend of a couple a couple of ingredients. Let's make
• 10:31
  Speaker 1: it a perfect recipe of AI adoption. So one, you
• 10:36
  Speaker 1: would need a decision clarity before AI. We need to
• 10:40
  Speaker 1: define the decision. If you ask it a vague question,
• 10:44
  Speaker 1: no model can help you. So let's make that clear, right, okay.
• 10:49
  Speaker 1: Second ingredient would be the data foundations. Not a perfect data,
• 10:53
  Speaker 1: but consistent, well understood data. Basically the source of truth
• 10:58
  Speaker 1: that AI can reference, that would be always updated, that
• 11:03
  Speaker 1: would always have the most truthful information. AI can handle noise,
• 11:08
  Speaker 1: no problem, but it cannot handle confusion. If you give
• 11:12
  Speaker 1: it two different data sources, it will start to mix
• 11:17
  Speaker 1: them together or make up assumptions that otherwise aren't present.
• 11:22
  Speaker 1: And third, I would say that's the biggest one is
• 11:25
  Speaker 1: the organizational mindset. The teams need to trust the recommendations,
• 11:31
  Speaker 1: not blindly, but with transparency. Say they can say, hey,
• 11:36
  Speaker 1: this decision was made with the help of AI, because
• 11:40
  Speaker 1: this is the portion where AI analyzed, this is where
• 11:43
  Speaker 1: we checked off. So I always have this transparency where
• 11:46
  Speaker 1: AI helps the human or human helps the AI, so
• 11:50
  Speaker 1: we can have this trace point system to look back
• 11:55
  Speaker 1: at all.
• 11:56
  Speaker 2: Right, So when AI is working well within this operation,
• 12:02
  Speaker 2: what does that success actually look like? Are there KPIs
• 12:08
  Speaker 2: that that shift decision making patterns changing? You know, what
• 12:17
  Speaker 2: does that look like for that organization?
• 12:20
  Speaker 1: That's a great point, actually, KPI is one of them.
• 12:24
  Speaker 1: KPI improvements usually come up with ten to thirty percent
• 12:29
  Speaker 1: improvement in their labor planning accuracy, uplift in throughput consistency,
• 12:34
  Speaker 1: reduced over time, better inventory return turns, fewer expedise, higher
• 12:42
  Speaker 1: service levels. Those are very nice predictable patterns that we
• 12:46
  Speaker 1: can put on a PowerPoint. But what about something that
• 12:50
  Speaker 1: can make the organization even better? I would go and
• 12:54
  Speaker 1: go ahead and say behavior shifts is what you can
• 12:58
  Speaker 1: measure your AI deployment success on. This is usually also
• 13:03
  Speaker 1: the first signal, so this is the easiest to gauge
• 13:07
  Speaker 1: and measure. As leaders come on board and discuss AI
• 13:11
  Speaker 1: into their organizations, For example, planners will stop exporting to
• 13:15
  Speaker 1: Excel every morning, so that's a big pain point that
• 13:20
  Speaker 1: planners all over the world have. Supervisors will walk in
• 13:25
  Speaker 1: with their prioritized action list instead of problems that they
• 13:28
  Speaker 1: would need to catch up on. Similarly, firefighting meetings will
• 13:33
  Speaker 1: get shorter, less frequent, less check ins. Leaders will start
• 13:38
  Speaker 1: planning what if scenarios instead of reconciling yesterday's task that
• 13:43
  Speaker 1: should have been done a week ago.
• 13:47
  Speaker 2: I know that one of the things that changed, at
• 13:49
  Speaker 2: least in our organization or in my department in our
• 13:54
  Speaker 2: organization is once we started utilizing AI tools on a
• 13:59
  Speaker 2: regular base, it almost became its own thing, where once
• 14:05
  Speaker 2: you started using it, you were coming up with more
• 14:08
  Speaker 2: ideas of how can I how else can I use this?
• 14:10
  Speaker 2: How can I do this again? How can I apply
• 14:12
  Speaker 2: this elsewhere? How can I make this work for this
• 14:15
  Speaker 2: other area? And for us in the marketing department, that
• 14:19
  Speaker 2: was kind of one of those key indicators of you
• 14:23
  Speaker 2: know this, this is working, this is taking off, We're
• 14:26
  Speaker 2: trusting some of the results, and you know, that's how
• 14:29
  Speaker 2: we knew that AI had taken hold in our business
• 14:33
  Speaker 2: unit was because we were then excited about using it
• 14:37
  Speaker 2: and trying to apply it in different places.
• 14:40
  Speaker 1: Right, right?
• 14:41
  Speaker 2: Do you find something similar when it comes to to
• 14:44
  Speaker 2: supply chain different activities within a facility, within within a network.
• 14:50
  Speaker 1: I would say, so the manufacturing client that comes to
• 14:54
  Speaker 1: my mind, how to schedule adherence problem? Too many surprises,
• 14:59
  Speaker 1: too much rework, So AI began identifying early warning indicators
• 15:04
  Speaker 1: from small, small anomalies in the upstream data to big,
• 15:10
  Speaker 1: big whale problems later down their pipeline. Suddenly the plan
• 15:15
  Speaker 1: manager went from why did this happen? To we saw
• 15:19
  Speaker 1: this coming and here's how we're addressing it. So AI
• 15:23
  Speaker 1: acts as sort of a therapist for many of the planners, saying, hey,
• 15:29
  Speaker 1: everything's fine, we're going to go through this together and
• 15:33
  Speaker 1: here's how we're going to fix it. From reactive to proactive,
• 15:37
  Speaker 1: from hindsight to actual foresight.
• 15:40
  Speaker 2: That sounds amazing. So four companies that are eager to
• 15:45
  Speaker 2: embed AI into some of their workflows is what's a
• 15:50
  Speaker 2: good starting point? Where would you even begin? There's so
• 15:53
  Speaker 2: many applications for it. Is there a single place where
• 15:59
  Speaker 2: organizations can say this is what we want to do.
• 16:02
  Speaker 1: That's a great question, because the best starting point is
• 16:05
  Speaker 1: always picking the high friction. I always love this book
• 16:10
  Speaker 1: by Brain Tracy, Eat the Frog in the Morning, or
• 16:14
  Speaker 1: many motivational speakers like to point to that direction as well.
• 16:18
  Speaker 1: I would say, pick one high friction and high value
• 16:21
  Speaker 1: decision that has a clear owner, has measurable KPIs, has
• 16:28
  Speaker 1: accessible data not just one time, but over the period
• 16:33
  Speaker 1: of a substantial amount of months, let's say six months
• 16:37
  Speaker 1: should be sustainable to avoid any seasonality issues, operational pain
• 16:44
  Speaker 1: today and the future and in the past. For example,
• 16:48
  Speaker 1: you can predict their labor needs. You can identify imminent
• 16:53
  Speaker 1: service risks, optimized for safety stock, and detect different abnormality
• 17:00
  Speaker 1: in their automation systems. Start with one workflow, not the
• 17:04
  Speaker 1: whole enterprise.
• 17:07
  Speaker 2: Don't try to boil the ocean exactly. Well, conversely, what
• 17:11
  Speaker 2: is one pitfall or misconception that they should watch out for,
• 17:16
  Speaker 2: like trying to boil the ocean.
• 17:18
  Speaker 1: I would say trying to boil the ocean is similar
• 17:23
  Speaker 1: to enterprise wide AI transformation immediately trying to combine different
• 17:28
  Speaker 1: CRMs together, trying to combine this task with sales task.
• 17:35
  Speaker 1: This leads to misalignment, slow value delivery load, the option
• 17:40
  Speaker 1: budget fatigue. So AI should deliver its first win within
• 17:44
  Speaker 1: the thirty to ninety days, not two years. So that's
• 17:48
  Speaker 1: a great, great news for many senior leaders who want
• 17:52
  Speaker 1: to adopt AI in their organization. So once people feel
• 17:56
  Speaker 1: the value, the scaling should come naturally.
• 18:00
  Speaker 2: That's that's a great measuring stick to say, you know,
• 18:03
  Speaker 2: if you're trying to do too much all at once.
• 18:05
  Speaker 2: You won't see the results this early, but you know,
• 18:08
  Speaker 2: in order to validate that your results are at least
• 18:14
  Speaker 2: somewhere in the scope of reasonable and not. You know
• 18:20
  Speaker 2: that the AI isn't coming up with with some fantasy
• 18:24
  Speaker 2: that you've got some reliable data that you've been working
• 18:27
  Speaker 2: with and you know, I think that the thirty to
• 18:30
  Speaker 2: sixty days is a reasonable amount of time to say
• 18:34
  Speaker 2: we've been working with this, we've been trying this, and
• 18:37
  Speaker 2: if you're not finding actionable results within that timeframe, okay,
• 18:41
  Speaker 2: you you haven't wasted an entire year on this project.
• 18:45
  Speaker 2: You can go back, you can tweak, you can test again.
• 18:47
  Speaker 1: Absolutely, in the grand scheme of things in the business world,
• 18:51
  Speaker 1: thirty two ninety days is usually the onboarding of new employees,
• 18:56
  Speaker 1: so you can always count AI as a new employee.
• 18:59
  Speaker 2: Ah, that's an interesting take. I like it. So in
• 19:03
  Speaker 2: one sentence, what is the biggest myth about AI capabilities
• 19:09
  Speaker 2: in supply chain that you wish more leaders would let
• 19:13
  Speaker 2: go of?
• 19:14
  Speaker 1: I would say the myth is that AI replaces human
• 19:20
  Speaker 1: judgment is quite prevalent in today's media, but in reality,
• 19:26
  Speaker 1: the organizations winning with AI are the ones that are
• 19:30
  Speaker 1: elevating their people, not replacing them. Kind of like our
• 19:35
  Speaker 1: organization at Kneebok, we're enabling the employees to use AI
• 19:40
  Speaker 1: in their everyday tasks to streamline the processes to make
• 19:45
  Speaker 1: them much more productive.
• 19:48
  Speaker 2: All right, So I think we can all agree that
• 19:52
  Speaker 2: the computers are not going to take over the world,
• 19:54
  Speaker 2: at least not this week, and that AI is just
• 19:59
  Speaker 2: an tool in our toolbox. We have lots of tools.
• 20:02
  Speaker 2: We have the old standby Excel, we have lots of
• 20:07
  Speaker 2: other tools available out there, but some of these AI
• 20:10
  Speaker 2: tools are fantastic tools to be able to have in
• 20:16
  Speaker 2: our back pocket, to utilize when it comes to our
• 20:19
  Speaker 2: organizations and hopefully move the needle faster or more efficiently.
• 20:26
  Speaker 1: Absolutely, I would agree with that there is always it's
• 20:31
  Speaker 1: always better to have multiple options than not at all.
• 20:34
  Speaker 1: And the more things you have in the tool belt
• 20:36
  Speaker 1: the better.
• 20:38
  Speaker 2: All right, Well, thank you to our listeners for tuning
• 20:41
  Speaker 2: into speaking of supply chain. If this conversation sparked a
• 20:45
  Speaker 2: new idea or touched on a challenge that you're navigating,
• 20:48
  Speaker 2: we'd love to hear from you. There are links in
• 20:50
  Speaker 2: the episode description for you to join the conversation online
• 20:53
  Speaker 2: and tune in next time as we look closer at
• 20:56
  Speaker 2: the fast changing dynamics of emerging markets. Growth opportunities are everywhere,
• 21:01
  Speaker 2: but so are the risks. Be sure to follow us
• 21:04
  Speaker 2: on your favorite podcast app so you don't miss out
• 21:07
  Speaker 2: on hearing how global design meets local execution.
• 21:16
  Speaker 3: You've been listening to Speaking of Supply Chain, a meboch podcast.
• 21:19
  Speaker 3: Keep connected with us by subscribing to the show in
• 21:22
  Speaker 3: your favorite podcast player. If you like what you've heard,
• 21:24
  Speaker 3: please rate the show that helps us to keep delivering
• 21:27
  Speaker 3: the latest in supply chain information. Thanks for listening, Until
• 21:31
  Speaker 3: next time.