Latest / Speaking of Supply Chain / Lightening the Load: AI's Lift in Logistics Routine Tasks

Transcript

Lightening the Load: AI's Lift in Logistics Routine Tasks · Speaking of Supply Chain

0:06
Speaker 1: You're listening to Speaking of Supply Chain, a meboch podcast.
0:10
Speaker 1: This is a show for logistics professionals looking to learn
0:13
Speaker 1: more about the latest innovations in supply chain. Each episode
0:16
Speaker 1: will feature a conversation on topics such as mitigating supply
0:19
Speaker 1: chain disruption and reducing risk, current automation trends, sustainability initiatives,
0:25
Speaker 1: and more. Let's dive right in.
0:31
Speaker 2: Hello, and welcome to Speaking of Supply Chain, where we
0:33
Speaker 2: explore trends, current events, and innovations impacting the logistics and
0:37
Speaker 2: supply chain industries. I'm your host, Ellen Wood. The supply
0:41
Speaker 2: chain industry, and really all industries right now, are in
0:44
Speaker 2: the midst of a major transformation with the increasing adoption
0:47
Speaker 2: of AI, it's automating processes, remote teams are collaborating across continents,
0:53
Speaker 2: and a global talent pool is within our reach. Today,
0:58
Speaker 2: we're going to explore how some of these forces are
1:00
Speaker 2: reshaping businesses, boosting efficiency, and creating new opportunities for those
1:04
Speaker 2: who are ready to embrace the power of artificial intelligence.
1:08
Speaker 2: Joining me today for our discussion is Michael Greenberg, founder
1:11
Speaker 2: and CEO of Third Brain Automation. Welcome Michael, Ellen.
1:15
Speaker 3: It's great to be here. Thanks for having me, of course.
1:18
Speaker 2: So let's get to our topic today, which is AI,
1:22
Speaker 2: and I want to talk about it in logistics, for
1:25
Speaker 2: supply chain and logistics. Can you give our listeners a
1:29
Speaker 2: clearer picture on how it's already impacting the business landscape,
1:34
Speaker 2: maybe specifically supply chain. What are some of the practical
1:37
Speaker 2: applications that you're seeing and that your company is providing.
1:41
Speaker 3: So I think the one we see that every company
1:46
Speaker 3: that we work with that has a logistics or supply
1:49
Speaker 3: chain component is putting into place is the simplest one,
1:54
Speaker 3: and that is just taking all of the emails and
1:58
Speaker 3: records and creating a little we call it a supply
2:02
Speaker 3: chain CRM, but in order to keep all their information
2:08
Speaker 3: organized and taking you know, oh, whenever an email comes
2:12
Speaker 3: into our inbox related to anything that is an active shipment,
2:17
Speaker 3: we want that email associated and we want somebody notified
2:20
Speaker 3: immediately and believe it or not understanding an email to
2:25
Speaker 3: route it like that is actually a great use case of.
2:29
Speaker 2: AI, absolutely because it's so easy to just parse the
2:32
Speaker 2: information that's in that data in that file and tell
2:36
Speaker 2: it to do something based on whatever criteria. I know
2:40
Speaker 2: that a lot of companies out there are experiencing AI
2:45
Speaker 2: from a different perspective, So more on the creative use.
2:49
Speaker 2: Make a picture fun in photoshop or in in design,
2:53
Speaker 2: I mean I play with that one myself. Help me
2:55
Speaker 2: write this article, Help me make this more interesting, help
2:58
Speaker 2: me make this engaging. Change the tone or the voice
3:01
Speaker 2: on something when it's generative AI. But when we're talking
3:05
Speaker 2: about businesses, we're talking about really just utilizing the data
3:09
Speaker 2: that they have on hand, and probably more than they
3:13
Speaker 2: even realize they have. I mean, we've got so much
3:16
Speaker 2: data accessible to us. Now, what are some of those
3:20
Speaker 2: challenges that supply chain professionals might face when it comes
3:24
Speaker 2: to adopting that? Is it an issue of not understanding
3:27
Speaker 2: what the possibilities are or misunderstanding how the use case
3:32
Speaker 2: is relevant.
3:33
Speaker 3: So I'm going to drop a I think a buzzword
3:38
Speaker 3: that we've all had an executive tell us we need
3:42
Speaker 3: and that some of us have and are lucky to have,
3:45
Speaker 3: and some of us want and don't have yet. That's
3:47
Speaker 3: a control.
3:48
Speaker 2: Tower, of course. Yes.
3:50
Speaker 3: So in order to actually have that control tower and
3:54
Speaker 3: in order to use AI on like the data you have,
3:59
Speaker 3: the the problem is normally the same one for both.
4:04
Speaker 3: And that's don't actually have infrastructure that allows all your
4:09
Speaker 3: data to be connected and stored in a single place,
4:14
Speaker 3: so that way you can build anything else on top
4:17
Speaker 3: of it. So for us, every time it's oh, we
4:22
Speaker 3: want to be able to track this metric or we
4:25
Speaker 3: want to be able to see these pieces combine, and
4:29
Speaker 3: we're going to go in and we're going to find out, Oh,
4:32
Speaker 3: your ERP doesn't talk with your CRM, or you have
4:36
Speaker 3: a specialized manufacturing tool that it doesn't talk with anything else.
4:43
Speaker 3: Whatever it is, you do not have. I guarantee a
4:46
Speaker 3: unified data layer, some warehouse, some data lake where we
4:51
Speaker 3: can actually combine the business information effectively.
4:55
Speaker 2: When we have that, how do they go about getting that.
4:58
Speaker 3: A lot of work, So it's a multi step process. First,
5:05
Speaker 3: we're going to document the way everything is set up
5:09
Speaker 3: now and the capabilities of existing tools, So that first
5:13
Speaker 3: capabilities audit is going to allow us to determine can
5:18
Speaker 3: we do this with the tools we have in place
5:20
Speaker 3: right now? If not, what are we able to do
5:24
Speaker 3: and what tools would have to change in order for
5:28
Speaker 3: us to make this possible. Once we know that, then
5:32
Speaker 3: we can design a solution for the company's specific stack
5:37
Speaker 3: and generally that's going to be either some sort of
5:42
Speaker 3: like airtable or no code, low code middleware solution where
5:47
Speaker 3: we're just getting everything unified and talking to each other
5:50
Speaker 3: using this tool, or it's going to be us, Oh,
5:54
Speaker 3: we actually have all the pieces in place. Everything's good,
5:58
Speaker 3: let's layer in some sort of data where housing solution
6:01
Speaker 3: and then let's go and build that AI analyst or
6:05
Speaker 3: other pieces on top. So it's that you got to
6:09
Speaker 3: build the foundation, and once we've got a foundation, then
6:12
Speaker 3: we can start stacking the individual digital technology components on
6:16
Speaker 3: top that actually make our work easier.
6:19
Speaker 2: So, speaking of that work, what types of use cases
6:24
Speaker 2: are being used? Let's talk about supply chain first. So
6:27
Speaker 2: I've heard of forecasting demand, optimizing routes, managing inventory, but
6:33
Speaker 2: we already have some tools for those things. What does
6:36
Speaker 2: AI bring to it that's more intelligent than what some
6:41
Speaker 2: of these individual tools might be able to do.
6:44
Speaker 3: Yeah, So the way to think about this is that
6:49
Speaker 3: AI is going to allow you to take all of
6:52
Speaker 3: these individual tools and combine them through one interface. First off,
7:00
Speaker 3: two is going to allow you to do more complex
7:04
Speaker 3: analysis on top. So let's not just forecast, but let's
7:11
Speaker 3: forecast these three scenarios. Right, and AI could say, oh,
7:16
Speaker 3: I need to do scenario one and then change these
7:20
Speaker 3: four variables and do scenario two, and you don't have
7:25
Speaker 3: to go through explaining all the various steps, so long
7:29
Speaker 3: as you have some sort of robust forecasting process in
7:32
Speaker 3: place or a tool that the A I can call
7:34
Speaker 3: upon to actually do that work. And I think this
7:38
Speaker 3: is what we're seeing in all white collar jobs at
7:41
Speaker 3: the moment, really is AI is coming in and it's
7:44
Speaker 3: replacing all of the rote junior tasks where we used
7:49
Speaker 3: to have somebody to run and put together these things.
7:53
Speaker 3: A properly built system with AI layered in, you no
7:57
Speaker 3: longer need that person doing that, and they can go
8:00
Speaker 3: be exploring a harder, tougher problem like what would it
8:04
Speaker 3: take for us to move our back office overseas?
8:07
Speaker 2: Or learning how to read the data that's being generated
8:11
Speaker 2: and understanding you know, you've got three scenarios, how do
8:14
Speaker 2: you decide which one makes the most sense. So this
8:18
Speaker 2: goes perfectly into my next point, which is remote teams,
8:22
Speaker 2: the talent pool, labor shortages, and now we're talking about
8:27
Speaker 2: kicking the interns out and using AI for all of
8:30
Speaker 2: the things that they use. All that's scut work. I
8:33
Speaker 2: don't think we're kicking no, no, no, they don't have this
8:35
Speaker 2: gut work anymore. Yeah, so it gives certain advantages, but
8:40
Speaker 2: it does take away a lot of that practice. And
8:44
Speaker 2: it's not that you couldn't still make an in turn
8:47
Speaker 2: or you know, a junior person go and do some
8:49
Speaker 2: of these things to learn and to have that experience.
8:53
Speaker 2: But when it comes to those mission critical business decisions,
8:57
Speaker 2: you don't have to wait for that to happen. You
8:59
Speaker 2: can have our done it and let them run the exercise.
9:02
Speaker 2: But what's it doing to our labor pool and how
9:06
Speaker 2: is it changing that landscape for those newer, younger professionals.
9:11
Speaker 3: In order for us to use AI, everything has to
9:15
Speaker 3: be digital. In order for everything to be digital, we
9:20
Speaker 3: can move a role remote. So I think if the
9:25
Speaker 3: job can be done by an AI now or at
9:28
Speaker 3: some point in the future, it is likely that that
9:32
Speaker 3: job can also be done anywhere in the world. There
9:35
Speaker 3: is very little advantage to that. In logistics, in supply chain,
9:41
Speaker 3: these are some of the few professions that actually have
9:44
Speaker 3: benefit to understanding the regionality and the geography. And so
9:49
Speaker 3: I think at least in supply chain, we're going to
9:52
Speaker 3: see teams spread out more across the supply chain and
9:58
Speaker 3: then have more of the co unication and reporting handled
10:02
Speaker 3: by these AI tools and automations in between, because that
10:07
Speaker 3: in turn will learn more and be more valuable to
10:10
Speaker 3: the company if they're working out of the warehouse and
10:13
Speaker 3: they can understand what's actually happening on the ground rather
10:17
Speaker 3: than being locked up in an office doing Excel all day.
10:20
Speaker 2: True. Well, and the supply chain is it's physical, it's
10:24
Speaker 2: products moving from A to B to C to D,
10:27
Speaker 2: and you have to be there to touch it, to
10:31
Speaker 2: move it, and all of those positions are never ever
10:34
Speaker 2: going to go away. I don't care how many autonomous
10:37
Speaker 2: vehicles we have. People have to be there. So I
10:42
Speaker 2: love that this industry has a little bit of a
10:44
Speaker 2: protection for that, that that there are always going to
10:48
Speaker 2: need human beings, human people, but that we're also embracing
10:52
Speaker 2: some of this newer technology and this newer availability of information.
10:57
Speaker 2: So what key steps should company take? You've already talked
11:01
Speaker 2: about getting a data a control tower, getting all of
11:04
Speaker 2: their data to talk to each other or all in
11:06
Speaker 2: one place. If the disparate systems don't talk to each other,
11:10
Speaker 2: what else do they need to do? To prepare for
11:14
Speaker 2: using AI to make decisions in their business. Once they
11:19
Speaker 2: have all the information available to them, all the data
11:22
Speaker 2: is available, how do they prepare for that, because that's
11:26
Speaker 2: going to be a major shift, a severe change to
11:29
Speaker 2: the way they go about doing business. What things should
11:32
Speaker 2: they look up for.
11:33
Speaker 3: Well, they should probably have a lot of flow charts
11:36
Speaker 3: being made. Another great intern task, you know, make flow
11:43
Speaker 3: charts of how we make every decision, how we do
11:46
Speaker 3: every process. Decisions in particular are important. We already have
11:51
Speaker 3: platforms we can deploy for customers that allow you to
11:55
Speaker 3: do decision intelligence drag and drop. If this, then that
11:59
Speaker 3: pull all of your data from every reporting tool in
12:03
Speaker 3: and build automated decision trees to manage processes. That is
12:10
Speaker 3: entirely possible now. It is only limited by your ability
12:14
Speaker 3: to actually keep up documentation and to know how the
12:18
Speaker 3: decision is really made, which luckily, by Jane, it's a
12:23
Speaker 3: little easier to understand a lot of these things because
12:26
Speaker 3: it's not like, oh, well it's because so and so
12:30
Speaker 3: is kind of squirrely. It's oh, no, shipment needs to
12:34
Speaker 3: get from A to B and it needs to be
12:37
Speaker 3: at B by time. Why these are the drivers that
12:40
Speaker 3: can do this thing. These are the prices, and AI
12:43
Speaker 3: can limit that to the final two options for a
12:45
Speaker 3: human to choose.
12:47
Speaker 2: And that's going to simplify so many things. I know
12:51
Speaker 2: that we're using AI and exploring how it can be
12:54
Speaker 2: used in our business in marketing. Honestly, it's a lot
12:58
Speaker 2: of the generative content creation and that is super exciting
13:04
Speaker 2: for me. I love trying it. I love playing with
13:06
Speaker 2: the generators and seeing what they're able to do. But also,
13:10
Speaker 2: as you said, the process automation and process mapping. I've
13:14
Speaker 2: done process maps for my own activities and what's the
13:18
Speaker 2: best way to go about doing this? These are the
13:20
Speaker 2: ten things I need to do, and this is all
13:22
Speaker 2: of the work that comes after that. So it's able
13:26
Speaker 2: to help me organize my own work and organize the
13:29
Speaker 2: things that I need to get done in such a
13:33
Speaker 2: quick way that then I have time to do them,
13:38
Speaker 2: which is kind of nice.
13:39
Speaker 3: Exactly. That's maybe you're bringing up or we're walking around.
13:44
Speaker 3: I think a key point, and that's AI is not
13:47
Speaker 3: replacing anyone's jobs at this point. It might be able
13:51
Speaker 3: to remove eighty percent of a task but still need
13:55
Speaker 3: a human decision either at the front to guide it
13:58
Speaker 3: or at the end to sort of supers and filter.
14:01
Speaker 3: And so we're not really doing one hundred percent AI anything.
14:04
Speaker 2: Yet, No, And I don't anticipate that we will get there.
14:08
Speaker 3: Yeah, not for anything like physical or anything really that
14:12
Speaker 3: important like content. I think we can get to one
14:15
Speaker 3: hundred percent AI tweets.
14:18
Speaker 2: I don't know. I was listening to something I saw.
14:20
Speaker 2: It was a clip of an interview Don Lemon and
14:24
Speaker 2: Neil deGrasse Tyson, and Neil deGrasse Tyson's stance was that
14:30
Speaker 2: AI will eventually break the Internet because there will be
14:33
Speaker 2: so much fake information, so much information that hasn't been verified,
14:37
Speaker 2: that's been created, that it is just going to bury
14:42
Speaker 2: itself in falsehood that no one's going to believe that anymore.
14:47
Speaker 2: It's just going to be I don't even know what,
14:50
Speaker 2: but completely unreliable for information. And I think that's an
14:55
Speaker 2: interesting point that you know, we've already got fake news,
14:59
Speaker 2: We've already got people putting out incorrect information, false information intentionally,
15:06
Speaker 2: or putting out information that is inflammatory. So I could
15:09
Speaker 2: see where in that case in your example, that it
15:14
Speaker 2: actually we move away from it and say, no, we
15:17
Speaker 2: don't want to do this with AI and make a
15:20
Speaker 2: safe space. I think.
15:23
Speaker 3: I know, at least on Facebook. There's a very famous
15:28
Speaker 3: example that came out earlier this year of shrimp Jesus.
15:32
Speaker 2: I haven't seen that.
15:34
Speaker 3: So, like, this is happening in Facebook groups where AI
15:37
Speaker 3: are generating images and other AI are responding to them,
15:40
Speaker 3: and then they're generating new images, and so it's caused
15:43
Speaker 3: this loop where they create more and more crazy images
15:47
Speaker 3: over time. And one of them was shrimp Jesus that
15:53
Speaker 3: made the rounds a bunch of articles. This was maybe
15:56
Speaker 3: three or four months ago. But I think the dead
15:59
Speaker 3: Internet is very true. It is happening, and we're seeing
16:03
Speaker 3: people move to podcasts and to newsletters and other forms
16:08
Speaker 3: of content that are harder to replicate as ways to
16:12
Speaker 3: avoid that. So I think that reaction is already taking place.
16:16
Speaker 2: What generative tools have you used or do you like
16:19
Speaker 2: to use? Do you use the creative generative tools or
16:22
Speaker 2: do you I mean, obviously you create AI tools custom
16:26
Speaker 2: So have you just built your own things or do
16:30
Speaker 2: you use some of the open AI tools?
16:32
Speaker 1: Oh?
16:33
Speaker 3: Yeah, So we are primarily a claud shop. Anthropics models
16:37
Speaker 3: are the ones we generally recommend and use first, and
16:41
Speaker 3: that's what we as a team uses like our day
16:43
Speaker 3: to day chatbot. We have it in Slack and then
16:46
Speaker 3: also the team's plan. It does not do images yet,
16:50
Speaker 3: just because we're not really an image heavy company. We
16:53
Speaker 3: do use Gemini, especially for workflows and for processing long
16:57
Speaker 3: videos and summarizing them or creating workflow documentation from video recordings.
17:03
Speaker 3: Gemini is very useful for that because it has perfect context.
17:07
Speaker 3: That's what I use with our clients. Most of them
17:11
Speaker 3: are on open Ai and so we use a ton
17:14
Speaker 3: of GPT four. If we're doing like fine tune models,
17:18
Speaker 3: we are generally using open AI's stack, unless a client
17:24
Speaker 3: would like us to use open source or we feel
17:27
Speaker 3: a need to use open source for that client, in
17:29
Speaker 3: which case we use Lama and we've done Mistroll once,
17:32
Speaker 3: but I think we're probably going to mostly stick with
17:34
Speaker 3: Lama for open source. So yes, I use them all.
17:40
Speaker 3: I like them all. We actually use a company called
17:43
Speaker 3: eeden Ai as an API vendor to They give you
17:49
Speaker 3: one API and then you can call all of the
17:52
Speaker 3: different AI models from the same API, and so that's
17:55
Speaker 3: very handy when you're trying to use many of them
17:58
Speaker 3: at once.
17:59
Speaker 2: So what do you think is going to be the
18:02
Speaker 2: next evolution? Of this information age. I think we're already
18:06
Speaker 2: at four point oh, what's five point zero?
18:09
Speaker 3: So I don't know what five point zero is except
18:12
Speaker 3: beyond being able to do math. I think that it's
18:17
Speaker 3: pretty clear that it's going to be able to do that.
18:20
Speaker 3: At least the way I'm thinking about it is this
18:23
Speaker 3: four point oh. The current models, they are the best interns.
18:26
Speaker 3: I've ever had no offense to any of my prior interns.
18:29
Speaker 3: But you guys, we're nowhere near as smart. You were
18:33
Speaker 3: definitely not as hard working as this twenty four hour robot.
18:37
Speaker 2: It's hard to compete with that.
18:39
Speaker 3: It's also better at Google. Oh, I use perplexity. I
18:44
Speaker 3: love perplexity. I should say that fantastic search engine.
18:48
Speaker 2: I'm not familiar with that one. I'll have to try it.
18:50
Speaker 3: Yeah, super fun. But back to the topic. Currently, we've
18:55
Speaker 3: got like the best interns ever. I expect the next
18:58
Speaker 3: version will be good enough to replace like most of
19:02
Speaker 3: my lower level team members. But and I think in
19:05
Speaker 3: both cases this a big butt. Like our limitation right
19:10
Speaker 3: now is not on the models, it's on the infrastructure.
19:14
Speaker 3: The model we have today, if you build it into
19:18
Speaker 3: a thinking loop and you have it go through a
19:20
Speaker 3: step by step process. And this is what a lot
19:23
Speaker 3: of people call like hard coded agents, or you might
19:27
Speaker 3: hear people call them agentic processes where you're taking the
19:32
Speaker 3: thought process of somebody and then you're writing code to
19:36
Speaker 3: mimic that thought process with the AI that allows you
19:41
Speaker 3: to extend the current technology orders of magnitude farther. So
19:46
Speaker 3: we can deploy marketing analysts, and we do, and we
19:50
Speaker 3: are starting to do this for supply chain as well.
19:53
Speaker 3: That can provide daily reports to clients across a wide
19:57
Speaker 3: variety of topics, and can provide notified reports based on
20:03
Speaker 3: various barriers. And that is just hard coding processes where
20:07
Speaker 3: it's going to check the metrics and if this number
20:10
Speaker 3: goes out of line, or if this shipment is overdue,
20:13
Speaker 3: or any of a million other things, it can notify,
20:16
Speaker 3: pull the other data together and create that report. And
20:20
Speaker 3: that's not something it can do off the shelf. That's
20:23
Speaker 3: not something you can just ask GPT four to do.
20:26
Speaker 3: But GPD four can do that right now. That is
20:29
Speaker 3: the kind of thing where I think we'll see a
20:31
Speaker 3: lot more of that in the near future, and that
20:34
Speaker 3: will allow us to remove even more of this work
20:37
Speaker 3: that we really don't feel like we should be doing.
20:41
Speaker 2: And you know what, that's that's an interesting thing because
20:43
Speaker 2: I saw something on Facebook recently and it was saying,
20:47
Speaker 2: I want AI to to do the things I don't
20:50
Speaker 2: want to do so that I have time to do
20:51
Speaker 2: the things that I do want to do. I want
20:54
Speaker 2: to write, I want to paint, I want to be creative.
20:57
Speaker 2: I want to do all these things. I don't want
20:59
Speaker 2: the I don't want the chat thot to do that
21:01
Speaker 2: for me. I want to do that, and I want
21:03
Speaker 2: the rest of my day to be easy so that
21:05
Speaker 2: I can And I think it's going to take the
21:08
Speaker 2: place of a lot of manual processes that we've had
21:11
Speaker 2: a person doing over the years. But just like any
21:15
Speaker 2: improvements in technology over generations, hundreds of years, every time
21:20
Speaker 2: there is a major step up, there's an outrage, there's
21:24
Speaker 2: a confusion, there's a misunderstanding of how this is going
21:28
Speaker 2: to benefit or how it's going to hurt. And then
21:33
Speaker 2: as time goes on and the technology is adopted, we adapt.
21:37
Speaker 2: Humans adapt to what is happening around them, and we're
21:41
Speaker 2: going to adapt to using AI as well. It is
21:45
Speaker 2: going to take over jobs. It's going to be your
21:46
Speaker 2: best intern but that's okay, because our interns are going
21:50
Speaker 2: to get better. They're going to adapt, they're going to
21:52
Speaker 2: do something new. They're going to learn the programming. Instead
21:56
Speaker 2: of learning how to do the task, they're going to
21:58
Speaker 2: learn how to create the thing it does the task.
22:01
Speaker 2: So I think that it does have the potential to
22:04
Speaker 2: be a game changer for society in general, to be
22:09
Speaker 2: able to free up that time that we want to
22:11
Speaker 2: have to not work. Yes, there are going to be
22:15
Speaker 2: people who want to have the career that they are
22:17
Speaker 2: all in. But for those of us that you have
22:19
Speaker 2: a career because you have to work, you have to eat,
22:22
Speaker 2: the lights need to be on. We want our jobs
22:25
Speaker 2: to be easier so that we can go home at
22:27
Speaker 2: the end of the day and enjoy life.
22:29
Speaker 3: I think you hit the nail on the head. My
22:31
Speaker 3: hope is that this will be the first technology that
22:34
Speaker 3: sees us have shorter working hours rather than longer.
22:39
Speaker 2: I like that prediction. I wholeheartedly hope it comes true.
22:43
Speaker 3: Yeah, I should say that's a hope, not a prediction.
22:47
Speaker 3: There is too much evidence to the contrary.
22:50
Speaker 2: I'm pretty sure it's not going to happen in my
22:52
Speaker 2: working lifetime, but perhaps for our children there.
22:56
Speaker 3: Is hope fingers crossed.
22:58
Speaker 2: All right, Michael, thank you so much for joining me
23:03
Speaker 2: today and talking about where AI is in our ecosystem,
23:08
Speaker 2: in our business. I want you So you've talked this
23:12
Speaker 2: entire time about supply chain, which I know my listeners
23:16
Speaker 2: deeply appreciate. But you mentioned something earlier about automating information
23:22
Speaker 2: in an industry where that's just not normal, and it
23:28
Speaker 2: was the construction industry. So tell us a little bit
23:33
Speaker 2: more about that, because that sounds fascinating to me.
23:36
Speaker 3: Well, construction is really project management and then making sure
23:41
Speaker 3: somebody's there every day and that things aren't being done
23:44
Speaker 3: like terribly wrong. And it's a lot of project management,
23:50
Speaker 3: I should say. You know, like people don't understand construction
23:54
Speaker 3: is ninety percent project management.
23:56
Speaker 2: Yes, but also the result, the product of construction is
24:02
Speaker 2: so artistic, so tactile. It's creation of something. It's not
24:07
Speaker 2: just moving something from A to B. It is building something.
24:13
Speaker 2: How does it work for that?
24:14
Speaker 3: Well, we don't really touch most of that. That's a
24:17
Speaker 3: big part of the key. So like we're not going
24:20
Speaker 3: to change the way somebody stacks bricks. There are companies
24:23
Speaker 3: doing that. You can go call Brett Adcock see if
24:27
Speaker 3: you can get yourself a figure two. They just open
24:29
Speaker 3: the announcements. That thing looks like it can stack bricks.
24:34
Speaker 3: We're not building a brickstacker, but it is pretty easy
24:39
Speaker 3: for an AI to look at a photo of concrete
24:41
Speaker 3: and tell you if there's cracks in it. And so
24:44
Speaker 3: if you're a concrete slab poring company and you want
24:48
Speaker 3: to make sure every one of your slabs has no
24:49
Speaker 3: cracks when you're done, or you want to make sure
24:52
Speaker 3: there's no air bubbles potentially before it's done, you can
24:57
Speaker 3: take a photo and an AI can look at that
25:00
Speaker 3: photo and it can tell you if it thinks there's
25:02
Speaker 3: anything there. And it's not always going to be right,
25:06
Speaker 3: but it can be eighty five ninety percent right without
25:08
Speaker 3: fine tuning. So it can do the same thing on
25:11
Speaker 3: safety violations. It can do the same thing on a
25:15
Speaker 3: lot of different steps in the quoting process when customers
25:19
Speaker 3: are sending in images and so it's at least in construction,
25:26
Speaker 3: the trick is finding out where the edges are, and
25:30
Speaker 3: those edges are mostly related to We can do some
25:34
Speaker 3: of the things that have to be done in the field,
25:36
Speaker 3: but you still need somebody to take the photos and
25:41
Speaker 3: a lot of it we just can't touch, and we've
25:45
Speaker 3: got to educate their team on these are the buttons
25:48
Speaker 3: you press to do this thing. This is how you
25:50
Speaker 3: take the photo, this is how you upload the report,
25:53
Speaker 3: and then the rest is automated. And so for them,
25:57
Speaker 3: they understand we have to do less paperwork, we have
26:00
Speaker 3: to do less other things if we follow this process,
26:03
Speaker 3: and then for the back office team for the ownership,
26:07
Speaker 3: they can improve their collections, they can improve their processes,
26:10
Speaker 3: and they can get more done.
26:12
Speaker 2: Interesting, I am so thrilled with all of the different
26:16
Speaker 2: areas where AI can do something. I mean, you were
26:19
Speaker 2: just talking about checking for imperfections, but what if it
26:23
Speaker 2: can do an analysis of those photos that the customer
26:26
Speaker 2: send in and generate a quote based on what the
26:29
Speaker 2: repairs are actually needed. Which, again, if it has a
26:33
Speaker 2: higher accuracy looking at those photos than a human would,
26:37
Speaker 2: you're going to get more accurate quotes to your clients.
26:40
Speaker 2: You're going to be able to tell them exactly what's wrong,
26:43
Speaker 2: and obviously you do need a human to follow up
26:46
Speaker 2: after that. You can't just rely on the AI parsing
26:51
Speaker 2: that information and that digital image. You have to have
26:53
Speaker 2: a person go out there and touch it and feel
26:55
Speaker 2: it and verify it. But if you can take out
27:00
Speaker 2: eighty percent of the pre work and the interpretation and analysis,
27:04
Speaker 2: that human brains are faulty at a lot of the time,
27:09
Speaker 2: or just our eyes can't see that level of detail
27:12
Speaker 2: that these digital images are able to and that the
27:16
Speaker 2: AI is able to read. So it's not that the
27:19
Speaker 2: person isn't trying, they're not knowledgeable, it's that our eyes
27:22
Speaker 2: are limited. And if it's able to generate that level
27:27
Speaker 2: of work, that's amazing.
27:30
Speaker 3: Yeah, we are edging more and more into the on
27:33
Speaker 3: site use cases one as the tools get more advanced.
27:37
Speaker 3: But I think our biggest barrier that we're finding is
27:40
Speaker 3: as the teams in the field get more advanced, we
27:43
Speaker 3: are able to show them how to use more tools.
27:46
Speaker 2: All right, So it's it's an evolution.
27:49
Speaker 3: Yeah, I mean I could see something like this and well,
27:53
Speaker 3: supply chain has you know, like codes to actually track
27:56
Speaker 3: materials already across wide dish instances. I also know that
28:02
Speaker 3: occasionally people keep the thing that you scan and they
28:05
Speaker 3: remove some other pieces as a way to avoid registering
28:13
Speaker 3: losses so they can resell those off market and things
28:16
Speaker 3: like that, and that is a case where oh, if
28:21
Speaker 3: we have a picture and we've done this with a
28:24
Speaker 3: client who owns some restaurants, they have a camera and
28:27
Speaker 3: they're walking and so they know what it should look
28:32
Speaker 3: like roughly and check and be like, oh, has anyone
28:37
Speaker 3: come entered and changed anything in the walk in the
28:40
Speaker 3: next period of time, which would allows them to track Oh,
28:44
Speaker 3: we can very quickly figure out if somebody came in
28:47
Speaker 3: and stole about your box of chicken tenders. We can
28:50
Speaker 3: very quickly figure out who did it when, and the
28:53
Speaker 3: AI is going to tell us the rough time period.
28:56
Speaker 3: You know, it's a lot of this like supervisor stuff
28:59
Speaker 3: is what we're finding in the field.
29:01
Speaker 2: Well, I appreciate that it is going to poach as
29:05
Speaker 2: many supervisory positions as it will interns. Excellent.
29:10
Speaker 3: Yeah, this is an equal opportunity technology.
29:13
Speaker 2: It will level the playing field, that is for certain. Well, Michael,
29:17
Speaker 2: thank you so much for joining us today. I really
29:20
Speaker 2: appreciate your time and your insight into this technology that
29:25
Speaker 2: is absolutely going to change the entire game for businesses
29:29
Speaker 2: across the spectrum, not just supply chain and third Brain
29:34
Speaker 2: Automation is the name of your company. But I had
29:37
Speaker 2: trouble finding it, and so I'm going to tell our
29:39
Speaker 2: listeners it's three r D, not third. The word three
29:43
Speaker 2: r D third brain.
29:46
Speaker 3: Fine, we own a bunch of the other domains too,
29:49
Speaker 3: but I haven't. I haven't found all the spelling variations
29:52
Speaker 3: yet to pick them up and forward them.
29:55
Speaker 2: Well, it'll get there, I'm sure. Thank you so much
29:57
Speaker 2: for joining us. It was wonderful have you for our
30:01
Speaker 2: listeners out there. If you liked this episode, please rate
30:05
Speaker 2: or review it on whatever podcast platform you prefer. Thank
30:09
Speaker 2: you for tuning in, and we will see you again
30:11
Speaker 2: next time.
30:16
Speaker 1: Meeboch Consulting is one of the largest and most globally
30:19
Speaker 1: recognized supply chain consulting, engineering, and advisory firms. For nearly
30:24
Speaker 1: fifty years, we've helped clients achieve supply chain excellence and
30:28
Speaker 1: sustainable competitive advantage across the entire spectrum of the supply
30:32
Speaker 1: chain by delivering improvements and innovation strategically, tactically, and digitally.
30:38
Speaker 1: To learn more, visit meeboch dot com. You've been listening
30:42
Speaker 1: to Speaking Up Supply Chain, a meboch podcast, keep connected
30:45
Speaker 1: with us by subscribing to the show in your favorite
30:48
Speaker 1: podcast player. If you like what you've heard, please rate
30:50
Speaker 1: the show that helps us to keep delivering the latest
30:53
Speaker 1: in supply chain information. Thanks for listening. Until next time.