Latest / Speaking of Supply Chain / Smarter Supply Chains: GenAI & Logistics

Transcript

Smarter Supply Chains: GenAI & Logistics · Speaking of Supply Chain

0:06
You're listening to Speaking of Supply Chain, a meboch podcast. This is a
0:10
show for logistics professionals looking to learn more about the latest innovations in supply chain.
0:16
Each episode will feature a conversation on topics such as mitigating supply chain disruption
0:20
and reducing risk, current automation trends, sustainability initiatives, and more. Let's
0:26
dive right in. Hello, and welcome to Speaking of Supply Chain, where
0:33
we explore trends, current events, and innovations impacting the logistics and supply chain
0:37
industries. I'm your host, Ellen Wood. Joining me today is Pete Gret,
0:42
founder and CEO of the Blackrock Group, and we are going to be
0:45
exploring some of the implications and uses of generative AI in the supply chain REALM.
0:51
Welcome, Pete, Hey y'all, thanks for having me. So let's
0:55
get into our topic today, which is generative AI supply chain. Now I
1:00
use generative AI, or I've played with generative AI in marketing. I mean
1:07
that implication is quite obvious. Everyone can see how that would work. What
1:11
are some of the tools that you have tried for AI? Yeah, so
1:17
my primary tool is really just open AI with chat GPT first to market.
1:22
It's a market leader. I know there's a multitude of tools out there,
1:25
like Claude, and there are some tools now that companies are coming up with
1:29
that even have the ability to kind of at a macro level, utilize different
1:34
genais so that you can see what kinds of responses you would get, and
1:38
also to do some preparative analysis so that any of the hallucinations or weird things
1:42
might be able to shelter out. My primary tool is chat GPT, so
1:49
I have used chat GPT as well and I do like it. I also
1:52
like Google's option which was barred and is now Gemini, and I like it
1:57
a lot. What are some of the things that you have done with chat
2:00
GPT that help in any sort of way. I know, for marketing,
2:07
I use it to help come up with some interesting ways of wording things,
2:12
or some you know, just to spark creativity. But when it comes to
2:16
business and especially supply chain, this is taking over the role of an individual,
2:23
and it does in marketing as well. You know, if I'm not
2:27
the one coming up with the creative text, if something else is generating it,
2:31
the question is, well, then is their purpose for me? And
2:35
there clearly is because anyone who's used these tools knows that it doesn't give you
2:39
exactly what you want on the first try. It's never going to So what
2:45
do you use it for? Yeah? I use it as much as I
2:47
can for everything. I think it's just like having a new employee, right,
2:52
okay, someone who can give advice, someone who can do menial tasks.
2:57
It really is transformative. Really view it as a collaborative tool. Right,
3:01
So if you take a prompt engineering course, and I highly recommend folks
3:06
do that, you know, you can take a really great one from Vanderbilt
3:07
on Coursera for free, will take you like less than ten hours, and
3:13
you will be so much more efficient about how to use generative AI. You
3:16
know, that's a big deal, but it really becomes a conversation. My
3:21
focus with a lot of the work we do with jen ai is to not
3:24
try and boil the ocean and replace all the humans with robots, because you
3:30
know, that's just probably not going to happen, and it's definitely not going
3:34
to happen anytime soon. It's a new technology and it has to mature,
3:38
and it's going to take lots of iterations. So start small, Start with
3:42
how do I make an individual's job better? How can they be more efficient?
3:46
Can they be twenty five percent more efficient? Can maybe fifty percent more
3:49
efficient? As you noted marketing? I use it in all of my marketing.
3:53
It does a lot of the automation pieces. It does a lot of
3:54
content generation to just kind of fill in right, So I it is almost
4:00
like a coloring book. I'm going to draw the lines and it's got a
4:02
color in and it does a lot of that kind of work. It does
4:08
things like proposals, stuff like that. I like that analogy, only I
4:12
use it the opposite way, where I have it kind of draw the lines
4:15
and the parameters, and then I fill in the details of the interesting things
4:19
that are relevant to the topic. But I haven't taken that course so on
4:25
the prompt engineering, and I know that that's such a huge aspect of this
4:30
is you have to know what to ask. It's not as easy as googling.
4:33
I think we're all pretty good at googling and finding the answers that we're
4:38
looking for at this point, but when you're trying to get it to generate
4:41
something for you, you really have to understand how the machine's algorithms are working,
4:46
and how to get it to respond to what you're asking it to do
4:50
in a very specific way. Absolutely right. I posted about this the other
4:55
day that it used to be that you were really value person if you have
5:00
the right answers. And I think we are transitioning to can you ask really
5:04
good questions? Because if you can ask really good questions, and you're going
5:09
to give a lot of really good answers. And that's really what the trumpt
5:12
engineering will help with and really increases not just the productivity of what you're doing,
5:16
but the reliability of what you're doing. That's how you get rid of
5:19
the hallucinations. That's how you get rid of the unreliable answers. So,
5:25
as a business owner, you're talking about, you know, productivity. How
5:29
do you determine how much work should be generated by that AI versus that live
5:33
person. Yeah, you should generate as much with AI as you possibly can.
5:38
I really do believe that, and I think Wang at Nvidio, the
5:43
CEO, has said this. You know, why are you teaching your kids
5:45
coding? Yeah, that's that's not going to be a skill of the future.
5:48
Well, there's still be coders to still be coders. But you know
5:51
who's going to be left really really great coders who use AI to do ten
5:58
x fifty x they could do by themselves. You know, he's not going
6:01
to be left marginal or poor coders. And I think that's going to be
6:05
true with a lot of jobs, because if you have a person who's really
6:10
skilled in their craft, they know how to use tools and they know how
6:14
to use tools for leverage to really do more than they could do on their
6:18
own. And I think that's what we're going to see. And you know,
6:20
that's really where my push is to go individualize it. It's so cheap,
6:26
it's so easy, it carries so little risk, and your reward is
6:30
like immediate. You could have an employee take a prompt engineering class on a
6:33
Monday, and by the end of the week they could literally be fifty percent
6:39
more productive than what they were. And you know, why go through a
6:43
large scale transformation project and try and like remap your supply chain and have it
6:47
make decisions now, just make everybody's job better. And you know, if
6:53
an organization at fifty percent of employees get fifty percent more productive in the span
6:58
of like a month or two, that's really transformational in and of itself.
7:02
That's true. So we're talking mainly about chat functions and text, but generative
7:10
AI also includes graphic and image, and I know how that works for marketing.
7:15
You know, obviously we can create much more interesting graphical displays, whether
7:21
it's for trade shows or you know, a picture that we're putting into a
7:26
brochure or on our website. But there are implications for this in production,
7:31
in distribution, you know, right, sizing cartons or boxes and having this
7:36
program tell you more about what it is that you're looking at and not just
7:43
verbalizing. So what things have you seen in your experience using the imagery capabilities
7:48
of some of these generative AI. Yeah, and that's a really good point,
7:53
right, So, especially from a business perspective, the user interface of
7:57
just a human being is really powerful, but it's rather limited. What you
8:01
can do with the API when you actually send it files is much more expansive.
8:07
So you not only have the capability to generate images, but you can
8:11
generate speech. You can do visual recognition, so you can send it an
8:16
image and have it do things with the image, right, And I think
8:20
a lot of that requires even more specificity in your design though, you have
8:24
to be really good for anybody who's played with any of the image generation.
8:30
You know, if you give it a vague prompt, you might get some
8:33
selling nuts. And you know, for instance, I know that it does
8:37
not particularly care to do text in an image very well. They not have
8:43
much luck with that currently. But it takes a lot of thought, right,
8:46
and that takes a little bit more work, and it takes a little
8:50
bit more planning on how you're going to do things. You know, this
8:52
is one of the things in supply chain that we tell our customers all the
8:54
time. Have a good, well defined process and well trained people before you
9:01
try and use an accelerator like technology. Because if you don't have those things
9:07
and you put the accelerator on, you're going to go the wrong direction faster,
9:11
yes, and that's just not good, or you're going to fail faster.
9:15
You are, you are, and you're going to get And this is
9:16
I think where we see some of the hype about you know, well,
9:20
it's hallucinating. It's giving me weird answers and historically inaccurate things. It's like,
9:24
well, did you have a good process and train your people well use
9:28
the tool like it was supposed to be used, right, was the foundation
9:31
there? Yeah, I'm not sure about all of those. Okay, So
9:35
what are some certain specific functions that you feel this has the quickest way to
9:43
get into supply chain? So we've talked about just reducing the effort from the
9:48
employees, and that can be a two sided coin. We can look at
9:50
that as either they can do twice as much work in the same amount of
9:54
time, or we can get to that point where you know, maybe we
9:58
can get by with a four day work week or something like that, where
10:01
we're able to spend more time in our lives rather than in our jobs.
10:07
You know, that could go two different ways. But what are some functions
10:09
that in supply chain we can use this for and see those quick wins?
10:13
Yeah, so I think a lot of that for me is you can start
10:18
with some of your kind of back office operations stuff. So in every company,
10:22
there are so many things that you have to do that don't necessarily add
10:26
a whole lot of value. Right, You've still got to fill out your
10:28
expense reports, You've still got to do your time cheap, You've still got
10:31
to do your approvals. There's this whole host of things that, boy,
10:35
if you had a really good add men, you would just give it to
10:39
them. Yes, And you know, that's a really easy quick win because
10:43
those things are really like known and tasks and steps and are really low risk
10:48
if something goes wrong. And I just start with those quick wins, easy
10:54
ways to buy some time back. And then even a lot of these things
10:56
like I don't do slide decks myself. I give the prompts and I work
11:01
interactively and I said, okay, let's work on the next line. Let's
11:05
go through this. Do you need to do swat analysis, you know?
11:09
Do you need to do proposals? Do you need to do other kinds of
11:11
things at work where you're needing to present to folks. All of that really
11:16
is really well done. If you do what's called a chain of thought,
11:20
right, this is how you would interact with another person. You wouldn't just
11:24
spew out everything all at once. It would be collaborative. You would go
11:28
back and forth and you'd be like, Okay, well I like that,
11:30
while I don't like this as much. We need to go this direction a
11:33
little bit more. We need to include this. That's how you can interact
11:35
with chap and so a lot of those tools are really easy to get some
11:41
value really quickly, and I have found that to be true. So when
11:45
someone is starting with one of these tools, especially, and I'm going to
11:48
focus on text because that's where I use it the most right now. And
11:52
it does come and go on whether I'm using text or visuals. It's seasonal
11:56
for me. But right now, when I'm using text, if something doesn't
12:01
sound right or if it's a little bit weird, you know, using those
12:05
was this helpful prompt that it always gives you. It always asks for feedback
12:09
every time it produces something, and it was this helpful? Was this good?
12:13
And you can say yes, but let's try something different over here or
12:18
no, that wasn't the direction I wanted it to go. Try this,
12:22
and you know, try again, but this, and it really does a
12:26
great job of taking that feedback and delivering you something else. So you can
12:31
work with that with that initial prompt of what you're trying to do and get
12:35
multiple responses back and hone down into what exactly you were looking for, and
12:41
it'll help you get there so much faster than trying to do it by yourself.
12:46
It really will. And there's a great feature in chad GPT where you
12:50
can make your own custom GPTs, so you can basically give it a memory.
12:54
Right, this is who I am, this is how I talk,
12:56
this is what I want to do. These are the tasks that I'm going
12:58
to look for you to do. Here's kind of explanations of how you would
13:01
do them, and it's really easy to answer. Pomorphize this right, So
13:05
then it's it seems like another person like Watson at IBM. It really is
13:11
right, And you know, sometimes I get a little frustrated with it because
13:15
so when I write copy, I'd be like, stop using rocket emojis.
13:18
You want to use rocket emojis every single time? Just stop And it's like
13:24
I'm sorry, I forgot again, and you know, stop. Everybody knows
13:28
it's AI when you use the rocket emojis, and that is why you have
13:31
to go back in and make a final pass through the text. But you
13:35
do. But you're absolutely right, And we're talking about with chat GPT.
13:39
Most people are utilizing these things with just general knowledge that's out there. But
13:43
you can really customize it and put in your own local data and have it
13:48
search your own local data for the insights that it needs in order to build
13:54
what you're asking it to rather than going out to the whole internet. Yeah,
13:58
it's really transformed in how you can even kind of synthesize knowledge. So
14:03
one of the things that I do for my own marketing copy is I keep
14:07
my reading list on CHRO. I see things that are interesting that I want
14:09
to comment on, I add them to my reading list, and then every
14:13
week on a Saturday morning, I will go through that reading list and I
14:16
will have a conversation with Chad GPT and I will say, go read this
14:18
article, synthesize it, give me the key points, here's how I want
14:22
to respond to it, write some ad copy for it, and it just
14:26
really does a great job of again heavy lifting. I'm not going to go
14:31
read all sixty articles, synthesize them into three bullet points, and then provide
14:35
analysis on them. I'm going to have it use my voice because I know
14:39
what I would say, and now it knows what I would say exactly.
14:43
It has learned how to be you with you not having to be there.
14:46
Yeah, exactly. And to your point, we still read over it.
14:50
We go okay, yeah, I like that. I don't like that.
14:52
Sometimes I'll have it updates. Sometimes I'll go in and make edits and we'll
14:54
get to a good spot. Have you ever had it come up with something
14:58
just completely off the rails when you really honestly expected it to come up with
15:03
something on point, absolutely right? So this happens especially as you get further
15:09
out. So like with Chad GPT four instead say three and a half,
15:13
or or chat GPD four turbo, where you'll say something and then it will
15:18
just say something with nonsense. I'll tell it to do something and I'll say,
15:22
well, I can't do that, and I'll say why and they'll say,
15:26
oh, I'm sorry, I didn't understand the question, and then it
15:28
does the task. So this again, this is why it's a collaborative tool,
15:33
and I kind of it really is like having a new employee. You're
15:35
gonna have to watch over the work a little bit. There's gonna be some
15:39
things that they know really well, and then you're gonna discover some things where
15:41
you're like, you must not have had to do that in your last job.
15:45
Okay, right, and you're gonna have to give a little more help.
15:48
But that's kind of that collaborative nature. And I think that's why some
15:52
businesses are not seeing the real value in this right now, because again they're
15:56
trying to do this thing that's huge and that really does replace a lot of
16:00
people instead of just going boy, if id to buy back fifty percent of
16:04
the time, all my employees would be on such a better spot. Because
16:08
most businesses today are affected by that tight labor market. It's not replacing any
16:14
people. We're not getting rid of any people. It just means instead of
16:17
having ten open recks, we'll have two open wrecks. Well, and I
16:21
was reading a little bit more about some ways that this can really drive value
16:26
in supply chain, specifically things like you know, AI is never going to
16:32
take over driving a truck cross country. It's never going to do that.
16:36
I know there are you know, driver lists features on cars, but that's
16:40
never going to happen in this supply chain. Transportation is going to require a
16:44
live people period. However, when it comes to the routing for these transportation
16:51
routes, which is going to make the most sense, which is going to
16:53
be the most cost effective, which is going to be the fastest, I
16:56
mean, each of those are different factors and and be completely different reasons to
17:02
choose that particular route, and you need the human there to make the determination
17:07
which is the most important right now? Is it more important to get it
17:10
there fast. Is it more important to get it there in the most efficient
17:15
travel route, or do we have to go around road construction or something else
17:21
roadblocks in the way, or is it most important to do this cost effectively.
17:26
And the human has to make that decision, but the GPT can help
17:30
you figure out what those options are to begin with. Absolutely right, And
17:34
this is a conversation we have with so many of our customers. We were
17:37
a modex a few weeks ago seeing all of the great technology and supply chain,
17:42
and everybody wants a robot. They don't know what they're going to use
17:45
it for, they don't know where it's going to go, but they know
17:48
they need to buy a robot because everybody else is buying a robot. And
17:51
I tell them, why why are you buying a robot? You know what
17:53
operational executional improvement does is like X. Decision improvement is like pen X make
18:03
better decisions. Once you make all of the best decisions, then worry about
18:07
how you can execute them. And your transportation analogy is spot on. That's
18:12
great. You picked a carrier that's got ninety eight percent reliability instead of ninety
18:17
six percent reliability. But you plan terrible routes, that's not going to matter.
18:22
Your executional excellence is just for naught because you turn the fine tune dial
18:27
instead of grabbing the big dial and trying to get closer to where your target
18:32
is and understanding all the options that are available. Because I think that's that's
18:37
one way that this can really to your point. Drive the efficiency is if
18:42
you tell your transportation director, find all of the variables here is it,
18:48
cost is it, time is it you know, gasoline or man hours?
18:52
What is the most important thing, and find all the options for it.
18:56
That person's going to be able to do all that. They're going to be
18:57
able to do the research, but take them a full day to give you
19:02
all ten options that you have versus putting this into a GPT, and it's
19:07
able to deliver all of those options much faster, and then that transportation person
19:12
doesn't have all of their time spent doing administrative work or research. They can
19:18
spend their time making the decisions and executing on them. It is, and
19:22
it really is like if you could afford all of the purpose built software that
19:26
would do each of those things, so I was talking with the customer in
19:30
the last few weeks about doing some work and seeing where it could provide some
19:36
recommendations in the software, right, and he said, well, you know,
19:37
maybe we can look at how these variables play together. And I was
19:41
like, well, why don't we just how to do the modeling for the
19:44
whole thing? Right? If you take the variables, you run out all
19:47
of the possible outcomes and you put probabilities to it. This is just probability
19:52
math and say, okay, well here's the three recommendations based on probabilities.
19:56
Because it can do that, and I don't have to have an advanced math
20:00
degree or special software for probabilities. I can just give it the right prompt
20:03
and give it the information from a file and it can do that like it
20:07
was purpose built to do that. And that's where I think so much of
20:11
this power comes in. It's like a Swiss army knife that really could do
20:15
all of those things. Well, right, you remember the Swiss army knife
20:18
was like marginal at everything. Yes, it was like not great at anything,
20:22
but mediocre and a lot of things. Well, imagine if it was
20:25
a Swiss army knife and it was graded everything. It was a great orkscrew.
20:30
It was a great knife. That's what you have. And to be
20:33
fair, you know, these open source or you know, widely available GPT
20:38
generators, they're fantastic, but they're not perfect. They're not going to be
20:41
able to come and deal with every company's data or it solves that they're trying
20:48
to find answers to. So that's where a lot of these very specialized algorithms
20:52
come into play in, you know, developing a tool that will analyze large
20:59
data lakes and find the answers that they're trying to pull out of that data
21:04
of you know, whether it's pick times in a distribution center, or we
21:10
already talked about transportation or production you know efficiency, all of those things that
21:15
can analyze all the data and give you different options for how you could optimize
21:19
this or how you could, you know, make this a little bit better.
21:23
But ultimately these tools are available, maybe not always out on Google or
21:30
chat GPT, but companies don't have to shy away from computer learning and that
21:36
ability to take control of their business from the inside. They don't always have
21:42
to, you know, go out and find new people or find talent.
21:47
They can use the talent they already have and just make that job that much
21:51
easier for those individuals. Absolutely, and I'm super excited about where we go
21:56
beyond jen AI. So I think of stacking these kind transformative technologies. Machine
22:02
learning in and of itself has tremendous opportunities, but a lot of time what
22:07
happens is is we don't have the right kind of input that would make that
22:11
really reliable. But what if we stacked GINAI on the front of machine learning
22:15
to help us kind of transform some of that data, understand those processes,
22:21
make the changes we would need to so that the mL would be really meaningful
22:25
and useful. And then I think all of that really needs to run on
22:29
a computational stack that is much more sustainable than what we have today. We
22:33
don't have the resources just to stay the growth and GENAI and so we need
22:38
quantum computers. And I know IBM has recently gotten to a thousand cubits and
22:42
at some point that's going to really unlock. You want to talk about transformation
22:48
qu on a computer is just on a whole different plane of existence. Really,
22:53
this is like sticks in rocks against the rocket show. I love that
22:57
you bring this up. I'm going to have to use this in my own
23:00
promotion of this episode when we start talking about it, because just for another
23:06
episode, I used a book that was given to my youngest son and it
23:10
was blockchain for Babies, and I also have quantum computing for babies, and
23:14
it makes it so easy to understand, you know, why this is important.
23:18
It's certainly not a reference book that any adult would use, but when
23:22
you're hearing these terms and thinking, oh gosh, that's way beyond my understanding.
23:27
I'm just a supply chain person, or I'm just a marketing person,
23:32
or I'm just an engineer that deals with industrial engineering. I'm not a computer
23:36
science person. But all of these things are so approachable with the technology that
23:41
we have today that the marketing person can understand the quantum computing that is going
23:47
on in these industries and in these business applications, and I just I think
23:52
it's so much fun. It's a great time to be in business in all
23:57
of these technologies and seeing all of them come together and shape the world that
24:03
we've got into something really amazing. It's a great time to It's so trite
24:07
to say it's a great time to be alive, but it is. This
24:11
is this makes work exciting for me? Well, and what we really try
24:15
and do, and this is I think the real secret to generative AI is
24:18
do only what only you can do, do meaningful work right. And this
24:25
is what excites me because this is what unlocks this for folks, that this
24:29
should be able to take the drudgery out of any job. Right. You
24:33
will do meaningful work because that's what you were design and creative for, is
24:38
to be a meaningful, productive person, not a drne. You're not a
24:42
drune. Be the special. You bring the values and the abilities that only
24:47
you can bring. And that's what's going to be so incredible to see because
24:51
I think you know, we talked about four day work weeks because people don't
24:55
want to be at work for five days. Well, what if you did
24:57
meaningful, exciting work, we're excited to get up every day and go do
25:02
your job. That is a different story, absolutely, and that is kind
25:06
of a different existence for most people. Forget about productivity, forget about what
25:10
it means for the economy. What does it mean for your life if you
25:12
enjoy your job every day? And I think that's just so exciting to see
25:18
or to have the option that even if you enjoy your job as much as
25:22
we're hoping that every individual gets the opportunity to do because it's meaningful. Even
25:26
if you do if you have a life where you want to have a four
25:30
day work, you can make that possible for yourself and still enjoy the work
25:33
that you're doing. But I know a lot of the pushback against these AI
25:38
tools and robotics, which we touched on, was that it's going to take
25:44
jobs away from humans, and I don't see that ever happening. I think
25:48
to your point, it's going to make the jobs that the humans are doing
25:51
much more impactful, much more value added. Rather than just being part of
25:56
a production line or stamping a thing, They're not going to be drones.
26:02
They're going to be value added to the equation. And I think you said
26:07
it best. It's just going to make everything more meaningful and allow people to
26:11
have that more meaningful work experience. Absolutely, and I saw this posting the
26:17
other day and it made absolute senses that at no time have any of these
26:19
technological improvements ever reduced the jobs in the market. Because new work is created,
26:26
new jobs are created, new openings are created, and that's what's going
26:29
to happen here. Plus the fact we have a tight labor market. You
26:32
know, we just need to fill all the jobs that we have. Exactly
26:36
to your earlier point, if we can have this takeover some of the administrative
26:40
some of these drudgery, easy tasks that just take up time in our day,
26:45
then we have more time for the actual value add things one hundred percent.
26:49
You know, there's a reason we have room buds, I have a
26:52
room book. That's why, right, because why would you spend time vacuuming?
26:56
What what value add is that for you? What special skill talent does
27:00
that employ of yours? Not so you know, but to your point,
27:03
even with a rumba, the rumba does not clean. It takes up a
27:08
lot of a lot of the work out of the equation, but you still
27:12
have to go back in there and get the corners. Because it's round and
27:15
rooms are square, you have to get under things. It doesn't dust,
27:19
there's not a dusting rumba. So there's still need for a human to be
27:26
in that equation. But let the robot do its job. Let the computer
27:30
do its job and make my job easier one hundred percent. And this is
27:34
just really a very visible way to see that. So Jenny, I is
27:40
going to do a lot of great things. It won't do everything, and
27:42
won't do all the good things that the proponent's saying, and won't do all
27:45
the bad things that the opponents say. It's going to be somewhere in the
27:48
middle, just like most things. All right, So tell me about a
27:53
time when it absolutely failed for you. It's for me. It's most of
27:57
the image generation. So I'm trying to do some content generation. I'll present
28:03
an image it's related to this, and it will try and put words in
28:07
there, or it tries to do a setting and I'm like, that doesn't
28:11
know, that's just not it. You need to try again. You misspelled
28:15
supply chain. Okay, we can't have that. And then you know,
28:19
after two or three times, if I can't get my prompt just right,
28:22
I'll be like, just take the words out just no no text. Yeah,
28:25
we'll try again later, no text, We'll just do something different.
28:29
All right. How about a time it was right on the money the first
28:32
time you tried. I am really surprised about its ability to do like deep
28:37
analysis and thinking with things like business proposals. And you say, hey,
28:42
I want the steps to write a book. I need to write a business
28:45
proposal for a book, and I want you to give me all the things
28:48
I need to fill in and ask me questions one at a time do you
28:51
have enough information? And then generate the proposal and it asks just real apprecient
28:56
questions that are spot on, and I answer them all and it up with
29:00
doom. Here you go. Here's a full proposal that meets all the requirements
29:04
that you had. It contains all the your information and requires no tweaks.
29:10
It's just perfect, fantastic. And that's just some deep thinking that you think,
29:15
Okay, wow, that's really that would have take I could have done
29:17
that, but that would have taken me like a couple of days, and
29:21
it did it in my fifteen minute conversation exactly. So how long have you
29:26
been using these generative AI tools that you got to that point where you can
29:30
either ask it and it's on the money, or how many failures have you
29:34
had throughout your process of trying to utilize this specifically for business. I know
29:41
I when I started, I was using it just for fun, just to
29:44
practice with it before I put it into a business application. But how long
29:48
have you been tweaking and playing with it? Yeah? So we've been doing
29:52
this for probably about two months. That's not long at all. So this
29:56
is a great example of this, right. So I started with kind of
29:59
like that personally for a few weeks and then it just the light bulb went
30:02
off and I thought, oh, this is it, right, this is
30:06
it. So I took all of our marketing. I took all of our
30:08
business and turned it towards Jenni. If you look at my profile on LinkedIn,
30:12
it says Jenai Evangelist. All of our outreaches on Jenai, all of
30:18
the podcasts we do are on Jenai, all of the new customers we're trying
30:21
to get are on Jeni. It is one hundred and fifty percent what we
30:26
are into because I believe it's that transformational. And it's not a linear path
30:30
where it's kind of going to slow over time. It's a logarithmic path,
30:33
right, so it grows by increments of ten factors of ten. It's a
30:37
big old curve like that, and what we do today will be absolute rubbish
30:42
in like three or four months, we'll just look back at that and go,
30:45
that's so childish. I can't believe that we even did that. Yeah,
30:48
it's just going to be that good, right, It just moves that
30:51
quickly. But to your point, it's not something to be just done by
30:53
itself. You have to guide it. It's a collaboration. It's a tool.
30:57
To me, it's just like having three of more extra people what I'm
31:00
doing, I can multiply my efforts to be so much more. It's a
31:04
lever right, It is a force multiplier, and we're just so one hundred
31:11
and fifty percent in on that. And I use it as a content multiplier.
31:15
So back to our coloring book analogy. You know, when I'm working
31:19
with colleagues and trying to develop content, I'll use the chat GPT to develop
31:25
that outline of what it is that we need to create, and then take
31:29
that to my colleagues and my experts and say, okay, what do you
31:33
think of this, and then bring it back because you know, they don't
31:36
always have time. They're busy doing project work. They're busy, you know,
31:40
executing on the things that require human brains to do. And then so
31:45
when I'm working with them to generate content, it is very collaborative, but
31:48
it's it's a three way process. But you're right, that maturity goes very
31:52
very fast. Of you know. Starting out in one of my first failures
31:56
was I was trying to write an invitation or a personal party. I was
32:00
having a girls' night, a summer event, and I wanted it to write
32:05
something that would get the individuals excited about this girl's night. It was Polynesian
32:10
themed, and so I just really wanted to get something interesting. And it
32:15
took me like four or five tries, and I was like, this is
32:17
not working at all for what I want. And then more recently, after
32:22
like you said, after you've practiced with it, after you've tried it,
32:25
after you've gone back and forth with it, and it learns who you are
32:29
and the way you speak and the way you want it to respond, then
32:34
I'm able to put something in and get something very quickly. Of you know,
32:37
this is the content piece that I'm trying to create, and this is
32:40
how I wanted to be presented, and these are some of the details that
32:45
it needs to have. And then get me a draft so that we can
32:49
work on it, and it immediately within five minutes, I'm done. I
32:54
don't have to sit and write and come up with all of this and do
32:58
research myself. It does it for me. So that maturity, you're absolutely
33:02
spot on, just super super fast. Yeah for sure. All right.
33:07
I think that brings us to the end of this episode. This has been
33:12
such a fun and interesting conversation. I know it's it's different based on your
33:16
business function for marketing, it's very different from business ownership and leadership, but
33:22
it's equally applicable, and so I think that our listeners are going to get
33:29
a lot out of today's episode. I'm glad that we could share this with
33:31
them because it's a tricky addition to any business to bring in a computer to
33:38
quote think for what it is that you're trying to do. Where can our
33:44
listeners reach you? LinkedIn is the best spot. I'm okay right on LinkedIn
33:50
and you can reach out to me there. We'll make sure to get that
33:52
link into our episode description. If our listeners out there have a suggestion for
33:57
a topic or would like to be a guest on the show'd love to hear
34:00
from you. But given that we're talking about AI this time, I don't
34:04
want you to email me at podcast at mebok dot com. I want you
34:07
to look for my LinkedIn post and use an AI chat tool. You can
34:12
go to chat GPT, or you can go to Gemini and just sign up.
34:15
It's free for the basic one, and use that chat to comment on
34:20
my post on what you'd like to hear about or that you think you would
34:23
be a great audition for this show to be a guest. That's what I
34:28
want our listeners to do if they want to participate this time. I'm going
34:30
to put something out on LinkedIn after this episode airs so we can get some
34:36
great interaction. As always, thank you everyone for listening to Speaking of Supply
34:40
Chain. If you enjoyed the show, please rate and review us on whatever
34:45
podcast platform you prefer, or on Apple Podcasts, Google or Spotify, and
34:49
be sure to tune in next time. Meeboch Consulting is one of the largest
34:58
and most globally rerecked nice supply chain consulting, engineering and advisory firms. For
35:04
nearly fifty years, we've helped clients achieve supply chain excellence and sustainable competitive advantage
35:09
across the entire spectrum of the supply chain by delivering improvements and innovation strategically,
35:15
tactically, and digitally. To learn more, visit meboch dot com. You've
35:21
been listening to Speaking of Supply Chain a meboch podcast. Keep connected with us
35:24
by subscribing to the show in your favorite podcast player. If you like what
35:29
you've heard, please rate the show that helps us to keep delivering the latest
35:32
in supply chain information. Thanks for listening. Until next time.