Latest / The ReForge Podcast / The Knowledge Is Retiring (with Matt and Colin) | ReForge 018

Transcript

The Knowledge Is Retiring (with Matt and Colin) | ReForge 018 · The ReForge Podcast

• 0:05
  Welcome to the Reforge podcast.
• 0:08
  I'm Matthew Chang, and we're here to explore all things AI, automation, and the American worker, proving that the American reindustrialization is possible, pivotal, and pressing.
• 0:19
  Thanks for joining us today.
• 0:21
  Let's get into the episode.
• 0:25
  Welcome to episode 18 of the Reforge podcast.
• 0:29
  We are now officially an adult podcast- Woo-hoo with all the rights and obligations that come with that now, so- Yeah.
• 0:37
  You mean Reforged Tattoo?
• 0:38
  We're real.
• 0:39
  I, I mean on our foreheads.
• 0:41
  I think that's the, uh, the- Yeah friend maybe.
• 0:43
  I'll, um, I'll punt on that though.
• 0:45
  But oh man, 18.
• 0:48
  It's cooking, cooking with, uh, gas now.
• 0:51
  So we have had some awesome conversations.
• 0:54
  Last few episodes covered some great things.
• 0:56
  Obviously, the one that we keep coming back to in one way, shape, or form or another is workforce.
• 1:02
  So today we're gonna take a little bit of a different angle on that, you know, on, on the workforce piece of it, and focus on the, the knowledge capture of it.
• 1:13
  Um, Matt, you and I have talked about it plenty.
• 1:16
  Silver tsunami is the most talked about, least understood, or least, you know, critical, critically thought about and conceived topic out there.
• 1:25
  Everybody likes to, you know, light their hair on fire about all these retirements, what's it gonna mean for the businesses.
• 1:32
  But there's something more critical that really isn't getting the attention I think you'd agree with, and that's, you know, we're talking decades and decades of experience, knowledge, knowhow, where the bodies are buried, all that good stuff with each person that retires.
• 1:48
  How are we capturing that?
• 1:50
  What do we do- Oh the ri- I mean, the risks are, I, I would love to see somebody quantify the risks.
• 1:56
  I don't even know where you would start with that because it's just you, you can't even measure how much these guys know.
• 2:01
  Everybody has their,"Well, how do you do that?
• 2:04
  Go ask Bob.
• 2:05
  Bob's done it for 30 years.
• 2:06
  He, he's the one who knows it." So I mean, w- where do we start with this?
• 2:11
  Well, let me, let me start by, um, I think our podcast today was informed by an account on X called Chief Engineer.
• 2:20
  Um, I banter with him sometimes.
• 2:23
  I call him the chief.
• 2:24
  So let me, um- It's not your throwaway.
• 2:26
  It's not your burner?
• 2:27
  Yeah.
• 2:27
  It's not my burner.
• 2:29
  It's actually a bigger account than mine, so it'd be funny if my, uh, my burner account was bigger than my main account.
• 2:34
  But here, here's what Chief has to say about it, right?
• 2:38
  So he, he's in here talking about engineering Wednesday, that's his theme.
• 2:44
  The great replacement is upon us, and I won't read the whole thing to you, I'll just kind of summarize it, but you can check it out.
• 2:50
  It's on his feed today What he's saying is, you know, he's talking about these 30-year keystone players in companies, and he's an...
• 2:59
  Like me, he's a professional engineer.
• 3:01
  And he, he makes the point that a lot of companies just don't have a plan because he's talking about people that have been at a, in a industry and at a specific company- Company man for 30, 40, and 55 years.
• 3:17
  And then they're stepping out, and the company realizes that they have no real succession plan, no documentation, no automation, nobody that knows the legacy systems, customers, and even the industry jargon.
• 3:29
  Those people step out, well, for a single company, that's problematic because now you're left with a customer support base that you can't serve.
• 3:38
  Or, you know, um, you're left with being unable to serve your industry'cause the guy or gal retired or ch- finally changed companies or had a heart attack.
• 3:51
  And then I, I kind of go on, I tagged you here.
• 3:53
  This is my, my, uh, repost, and I, I just wanted to frame it a little more simply than he did.
• 4:00
  It takes 15 years to make a good engineer, right?
• 4:03
  And this country, the data from 2022 was that we were, in civil engineering only, which is like concrete roads, bridges, structures, ports, we were 88,000 civil engineers short of being able to maintain infrastructure built out in this country.
• 4:21
  Rail lines, you know, levees.
• 4:23
  Wh- wh- which on its surface is bad- Mm-hmm but the fact that our infrastructure gets Ds and failing grades across the board to begin with means it's so much worse- Oh, yeah arguably if we can't even hit that 88.
• 4:37
  That is, that is a scary number.
• 4:39
  That's for infrastructure engineering.
• 4:40
  Yeah.
• 4:41
  Now let's, let's go to, you know, mechanical, electrical, um, aerospace, rocket engineering, right?
• 4:50
  All the different types...
• 4:51
  Automotive engineering.
• 4:53
  I'd say easy estimate, we're probably 500,000 short as a country, just like we're talking about a multi-million workforce shortage in advanced manufacturing.
• 5:05
  You know, so...
• 5:07
  And I, and I take them through the timeline bec- people will say,"15 years?
• 5:10
  That's crazy.
• 5:11
  This is not like a single presidency or a single DC policy is gonna fix this." Here's the simple timeline.
• 5:20
  So I have a high school senior, okay?
• 5:22
  That high school senior who has worked hard enough to actually get into engineering school, which you did, Colin, and so did I, uh, but it's not most of the high school seniors are getting into engineering school.
• 5:35
  Then they have to go to four to five years of college.
• 5:38
  I went five, but I did get a co-op degree.
• 5:40
  I also went, you know, went to a lot of parties.
• 5:44
  Then it takes five years of apprenticeship to become a professional engineer.
• 5:49
  And, and just for the audience, PE stands for professional engineer.
• 5:52
  It does not stand for private equity.
• 5:54
  I'm tired of, uh, finance bros- Or physical education.
• 5:57
  You know?
• 5:57
  Shout, shout out to the PE teachers.
• 6:00
  PE, I'm okay with PE teachers, that's cool.
• 6:02
  Uh, as long as they still climb the rope, right?
• 6:05
  Yeah.
• 6:05
  And then, and then another five years working in the trade as, you know, a becoming increasingly more independent licensed engineer.
• 6:15
  By that point, now you're good, and it's 15 years in the making, right?
• 6:21
  And that's if you start right now.
• 6:23
  Uh- So we're talking 2041.
• 6:26
  Yeah.
• 6:26
  Journeyman electrician, how long does that take for an electrician to become very good?
• 6:31
  Maybe not 15 years, but it's not, it's not less than five, right?
• 6:35
  No.
• 6:35
  I mean, mo- most apprenticeships are gonna be, you know, four or five years from, you know, basically hands, new hands to all the way up to being full member and union type thing.
• 6:44
  So you're at five years there, plus, you know, you've gotta get reps in after that, so five, seven, maybe 10 years.
• 6:51
  Yeah.
• 6:51
  I mean, it's, it's still not- Yeah, a good- I mean, it's better, but not much a good electrician knows systems and code like the back of their hand, and that takes a long time to learn.
• 7:03
  So I wanted to just kind of- It's scary give you guys a few.
• 7:06
  Yeah.
• 7:07
  And, and it's scary, but it's also, it's called a career, right?
• 7:09
  It's like these are skills- Oh, absolutely.
• 7:11
  Opportunity's huge that we have under-invested in as a country, we have de-emphasized them because of offshoring and because of promoting, let's say, kind of artificial things like the internet economy and, and social media.
• 7:28
  You know- To- how, how much talent did that soak up?
• 7:31
  You know, SAS companies, man, we love to dunk on them all of a sudden.
• 7:34
  I've been dunking on- There's some brilliant people, though.
• 7:36
  I mean, there, there are some- Oh, look- phenomenally, phenomenal people that, you know, decided that was the best, you know, course for, you know, economic returns for them.
• 7:44
  That was the way they wanted to go, get that big, get, build their bag.
• 7:47
  Imagine- But they could have done- Do a thought exercise some cool engineering stuff.
• 7:50
  Do a thought exercise.
• 7:51
  Take all of the engineering grads that went into quant, went into something like hedge funds or, you know, venture capital, private equity, where they use that math brain of theirs to attack spreadsheets.
• 8:06
  Take all those people and then take all of the, I'll call it software and Silicon Valley hype type companies that have come up and, and down in the past, you know, two decades, and all the people in there that are, went to engineering school and are technically qualified to do real engineering, and what if you had all of those people pursuing a traditional path in engineering?
• 8:27
  I'd argue that, that we really wouldn't be talking about an engineering shortage right now.
• 8:32
  We'd have the capacity to build more ports, bridges, rocket ships, railroads- Hands down skyscrapers, housing, right?
• 8:39
  Electrical ge- grid upgrades, um- Not even to mention the innovations in each of those spaces, just with having some of the best and brightest minds focusing on how we solve these problems better, faster- Yeah.
• 8:51
  And then- stronger, Daft Punk, whatever I won't tell you that all engineers are created equal.
• 8:56
  Just'cause you carry a PE behind your name, it's like carrying MD behind your name.
• 9:00
  Hopefully everybody in the audience has, has had an experience with a doctor that just didn't have all the lights on.
• 9:06
  Yeah, you went to med school.
• 9:07
  Yeah, you got licensed as a medical doctor.
• 9:10
  But- Yeah, had fun in the Caribbean for a few years and, you know.
• 9:12
  There, there's a reason why second opinions matter, right?
• 9:15
  And, um, and in engineering it's the same.
• 9:18
  We call, instead of second opinion, we call it peer review.
• 9:21
  We always do our designs in a way that somebody else could look at it.
• 9:24
  Well, I could tell you this.
• 9:26
  A lot of the engineering that I see done by other people in other companies, not very good, right?
• 9:33
  It's not creative.
• 9:34
  It's not insightful.
• 9:34
  It's wasteful.
• 9:35
  It could be overdone or underdone, so you're either designing an unsafe system or under- underperforming system if it's underdone.
• 9:43
  And then if it's overdone, you're just wasting the client's money, whether it be taxpayer money- I mean, you gotta...
• 9:47
  You, you have the whole that's how we've always done it phenomenon inside- That's it a lot of these firms.
• 9:51
  That's it.
• 9:52
  So- If just engineering takes 15 years to where...
• 9:57
  And, and quite honestly, that's what it takes to get to work at my company.
• 10:00
  So the person that wants to come to work at Chang Robotics, that 15 years already needs to be in place before we can even consider you for a, say a manager level role, a mid-level role.
• 10:11
  Um, otherwise you're going into entry level, right?
• 10:14
  Even if you're, think you're good.
• 10:16
  Oh, ex-Amazon, you worked there for four years.
• 10:18
  Great.
• 10:19
  Entry level, right?
• 10:20
  Because I would have to teach you everything.
• 10:23
  Um, so y- you know, we iden- we identified electricians could be out seven, eight years before they start to get really good and, and can really operate as, as a true leader on, say, a job site, right?
• 10:35
  And, you know, th- this goes across the entire industry.
• 10:39
  We're just talking about engineers and electricians as profiles.
• 10:42
  What about the welders?
• 10:43
  What about the machinists?
• 10:44
  I mean, you get a new guy...
• 10:46
  I remember when I started, when, when I was learning to weld, it's, you know, it, it is not pretty first few times.
• 10:54
  Until you learn how to lay a bead, until you how to mixer, mix the gas right, the right feed speed, all of that, it looks rough.
• 11:00
  Even once you do learn how to get that stuff down right, it takes a long time before it looks like you're stacking dimes.
• 11:07
  You've got to put in the reps.
• 11:09
  You have to do it.
• 11:10
  You've, you learn the tips and tricks from other people who learn them from somebody else.
• 11:15
  It- it's knowledge passed down.
• 11:17
  If you go into any machine shop, that, there's a guy who, you know, he hears that certain frequency and knows that that job he was running, that there's chatter on it, that they've gotta adjust the feed speed or spindle speed.
• 11:30
  He knows exactly what it is based on.
• 11:32
  He knows that it's, you know, 70, 75 aluminum in there, and okay, that's the job I'm running.
• 11:39
  We've gotta adjust something.
• 11:41
  That's not...
• 11:42
  It, it is something that you can learn But it takes time because somebody has to teach you.
• 11:48
  That's right.
• 11:48
  And that's where I think what, what's scary is the guys who- and gals who ha- know that stuff, who can be teaching it, they're the ones that are walking out the door, like you said.
• 12:00
  Yeah.
• 12:00
  And meanwhile, we have infrastructure report cards from across the country.
• 12:04
  Our airports need to be modernized, you know, certainly in both the infrastructure, but also the air traffic control.
• 12:11
  Oof.
• 12:12
  That has come up repeatedly that we're underutilized on air traffic capacity because we're still doing things the old way.
• 12:17
  We don't have modern equipment systems.
• 12:20
  Who's gonna design that?
• 12:21
  Who's gonna manufacture the product?
• 12:22
  Who's gonna install it?
• 12:24
  Who's gonna train the, the, uh, air traffic controllers?
• 12:27
  The, the training is e- How do we do that?
• 12:29
  Yeah.
• 12:29
  The, uh- It's scary.
• 12:31
  The bridges in our country, you know, it's embarrassing every time we have a bridge fall down and, and almost every year we have a headline about a bridge coming down.
• 12:39
  I think nationally we're somewhere around a D on our bridge scorecard.
• 12:43
  Pennsylvania, I mean, it is always, always, always, always a campaign issue you hear come up.
• 12:49
  I mean, years ago, you know, the...
• 12:51
  You'd drive down, you know, a highway and you could see a big sign.
• 12:54
  This is back when Governor Corbett, so was it two, two governors ago,"Please save our bridge, bridges." I, I mean, they're literally, you know, Midwest, you know, East Coast type thing, where you have this just legacy of heavy industry and stuff that post, post is industrial now.
• 13:14
  I mean, that infrastructure is literally falling apart.
• 13:17
  Florida at least, you know, Florida, Texas, places like that- Mm-hmm where they've had the influx of residents, they've had, you know, better budgeting, um, less, uh, less, uh, liabilities that they've had to fund through the nose.
• 13:30
  They've actually invested in a lot of that infrastructure, and you're seeing that, why you have so many jobs and companies relocating there.
• 13:37
  Yeah, I mean- It's, it's a major, major issue I live in a bridge city.
• 13:41
  You know, you can't get around Jacksonville without going over a bunch of bridges.
• 13:45
  We have water everywhere.
• 13:46
  And, and people don't realize that most bridges don't actually cross water, they cross, like, other highways or something.
• 13:51
  But if we just kinda zoom in on that, you know, we're operating...
• 13:55
  Our main street bridge, US 1, Highway 1 for the United States, runs through downtown Jacksonville.
• 14:01
  It's our main street.
• 14:03
  And it's got a 100-year-old steel drawbridge on it that's still in operations.
• 14:08
  It goes up and down several times every day.
• 14:10
  It's 100 years old, but it takes an intensive amount of maintenance to keep that bridge up and running.
• 14:16
  You know, the lane closures on it are common, repainting, refurbishing the, the mechanical parts that operate the drawbridge.
• 14:23
  It's not a simple task, and it still has to be able to communicate with boats so that it knows when to go up and let the boats pass under.
• 14:29
  It's a bit of a Frankenstein by now.
• 14:32
  Yeah.
• 14:32
  Yeah.
• 14:33
  And, and so, yeah, it's a lot of work to maintain your existing infrastructure.
• 14:36
  You know, we touched, we did airports, um, our roads across the nation, dams and levees, right?
• 14:43
  And that type of stuff doesn't matter until it fails, right?
• 14:47
  Yeah.
• 14:47
  Here, here's a trivia question.
• 14:48
  No- nobody likes to pay attention to it until it's too late, until it starts, you know, actually affecting people who don't have to look at it every day.
• 14:56
  So I'll prime you on this.
• 14:58
  The fir- the, the most, the most, um, I'll say, violent failure of a levee system is kind of still fresh in some of our minds, at least for people Colin, your, and I's age, and that was, um, the levee system- Katrina in New Orleans post-Katrina, right?
• 15:14
  So a hurricane- As a, as a native Houstonian, we, we saw that firsthand.
• 15:17
  Yeah.
• 15:18
  As a...
• 15:18
  Yes, you did.
• 15:19
  You took a lot of, a lot of refugees in to help feed them and take care of them in the interim,'cause it was, it was bad.
• 15:26
  So that's, you know, one hurricane, and the subsequent flooding knocks out a significant levee.
• 15:31
  It defends one of our most important cultural cities, right?
• 15:35
  D- would you like to guess what number two...
• 15:37
  And that's been repaired.
• 15:38
  That was the largest, uh, Army Corps of Civil Engineers project ever, was the repair of the New Orleans levee system.
• 15:45
  Lake Pontchartrain.
• 15:46
  Would, would you like to guess?
• 15:48
  Let's see if the audience can do it.
• 15:50
  What is the number two, and now it's number one, most at-risk city in the nation based on levees?
• 15:58
  New York.
• 15:59
  That's a pretty good guess.
• 16:01
  New York's actually blessed with how it sits.
• 16:03
  You know, Manhattan in general sits up pretty high.
• 16:06
  Uh, but we'll- I must have been watching too many, uh, too many apocalypse films.
• 16:10
  Yeah, yeah.
• 16:10
  It makes for good Hollywood.
• 16:12
  It is Sacramento, California.
• 16:15
  Ooh.
• 16:16
  Sacramento has a D or lower on every levee surrounding the city.
• 16:20
  It's a good thing- And, yeah that isn't the capital of a major, you know, economic power in the US, right?
• 16:26
  Yeah, the, the fifth or sixth largest economy in the world.
• 16:28
  Yeah.
• 16:29
  And, and, um, yeah, if you go to Sacramento on a normal day, the river's low.
• 16:34
  But what happens every spring is that snow melt comes, or they get one of these crazy weather systems, like the El Niños or La Niñas.
• 16:40
  And yeah, Sacramento's at serious risk of repeating what happened in New Orleans, and I think every, all the levee system around there is a, is a D or lower in terms of scorecard.
• 16:51
  So as a nation, it's kind of an imperative, and the infrastructure is just the foundation that allows us to build new things, right?
• 16:58
  If we don't have roads, bridges, airports, seaports, then we're not- You can't move things.
• 17:03
  You can't sell things.
• 17:03
  You can't sell things.
• 17:04
  Yeah.
• 17:04
  You're not making money, and you're a, you are a SOL.
• 17:07
  Yeah, you're not...
• 17:08
  If, if you don't have infrastructure, you're not getting a new manufacturing plant that we're pushing for, right?
• 17:12
  It's gotta be c- able to be connected via high- highways or roads or ports or what have you, and you gotta have the electrical capacity to connect the new factory to begin with.
• 17:21
  So this kind of infrastructure problem, the reason I started there, is it's kind of foundational.
• 17:26
  None of the stuff we're trying to do in the economy works without the basic infrastructure.
• 17:30
  Yeah.
• 17:30
  You gotta have, you gotta have fresh water, clean water, you know, good power.
• 17:35
  We've, we spent a lot of time on that recently, and you gotta have the ability to move things around in seaports, airports, and highways and railroads.
• 17:42
  So the, um, where I'm at with knowledge capture is I, I think for the most part, I feel like I called the AI bubble, and that the...
• 17:53
  You know, when we first saw ChatGPT, we were obviously all impressed But the, uh, this idea that it's going to AGI and the singularity, I was incredibly skeptical from day one, and I'm like,"I don't see it.
• 18:07
  This just looks like a better chat, you know, search engine." How many Rs are in the word raspberry?
• 18:13
  And they're like- I think four"Oh, but you can clone websites." And I'm like,"Yeah, I don't care,"'cause it was never in SaaS, right?
• 18:18
  Like, cloning a website doesn't add value to me.
• 18:22
  Yes, I'll do it in the future.
• 18:23
  I'll build my own apps and I'll, I'll, uh, create- But that's not core knowledge.
• 18:27
  Correct.
• 18:27
  That's, that's a, a feature, if you will.
• 18:30
  Yeah, that's not creating new, that's copying old, right?
• 18:33
  And so what...
• 18:34
  Where I think AI actually has a huge opportunity is in knowledge capture, right?
• 18:40
  Is- 100% agree are we, are we getting the thing, the, the deep technical knowledge from these experts?
• 18:47
  Uh, Chief in his post, he calls them technical directors.
• 18:50
  We actually use that title at my company.
• 18:52
  It's a very, it's a very seldom used title, kind of like a four-star general in the Marine Corps, right?
• 18:59
  It's, like, very rare and, uh, and it's only bestowed on the people that have the greatest capabilities.
• 19:04
  It's also for a short time.
• 19:05
  To be a technical director at our company, it's project or program specific.
• 19:10
  It's for a set period of time.
• 19:11
  Uh, and then after that you get demoted back to your regular job, right?
• 19:15
  So it's a high calling, and it's a very hard pinnacle to reach.
• 19:19
  What, what a lot of companies, a lot of industries are facing is the retirement or the moving on of these technical directors, the people that have that top 1% technical ability to be able to direct all the other technical people around them, all the engineers, all the programmers, uh, everybody.
• 19:37
  Um, and so I think AI, if companies are smart and, and if we as a society are smart, can really help in that knowledge capture from these elite top 1% technical people, um, and get it in a state to where it can be passed down, and I have a perfect fresh example for you.
• 19:58
  So- Let's hear it I would say 50% of the capacity of Chang Robotics, as a reminder for the audience, we do engineering and then build of advanced robotic systems.
• 20:10
  I'd say 50% of our time this year is gonna be spent in rebuilding the metals industry in this country.
• 20:18
  So that's a hot topic.
• 20:19
  We don't...
• 20:20
  We used to make all the metal for the whole world.
• 20:22
  We, now we don't.
• 20:23
  We've outsourced those industries to other nations where they would provide free power to the manufacturers.
• 20:31
  Um, but to find the technical expertise to design a new plant, which is what we're doing, we're designing new capacity for America in the metals industry for several companies, um, we have had to pull people out of retirement Like,'cause no one in America knows how to design a metal plant anymore.
• 20:53
  So we've had to pull people out of retirement.
• 20:55
  We've had to go deep into university Rolodexes and get people that are, that are already on their way out.
• 21:01
  You, you gotta dig deep for it.
• 21:03
  It's, it's not easy.
• 21:05
  And pro- a lot of people who've cracked down are dead.
• 21:08
  'Cause somebody'll say,"Oh yeah, I used to work with Frank on this.
• 21:10
  He was just magical around this chemistry, around this electrical process." We'll go looking for Frank, and Frank passed away four years ago, right?
• 21:20
  And then all- No, no more Frank he's gone.
• 21:23
  And so we have gone so far as to source SMEs from former Soviet Bloc countries that are also in retirement and will contribute their insights back to us.
• 21:34
  That's how far we're reaching to find people is retirement, universities, people that are fading out, and then people from other countries that have, have, uh, repatriated after their career.
• 21:47
  And it's been absolutely fascinating to see, like, we're this close.
• 21:54
  If we had decided to be- rebuild metal industry in this country 10 years from now, it would not be possible because all that technical director knowledge is gone.
• 22:02
  Gone.
• 22:02
  Right?
• 22:03
  So I think we're probably just in time to be able to get it done.
• 22:06
  I'm, I'm very optimistic about it.
• 22:08
  One of our friends on, on, uh, X, uh, AJ Cortez, he's doing this in advanced chemicals and precursors.
• 22:15
  Huge.
• 22:16
  How, how deep down the rabbit hole do you think he's going to find experts that can help him with the code?
• 22:21
  I, I can't even imagine.
• 22:23
  I mean, it's just...
• 22:24
  Where, where, it, it isn't...
• 22:27
  Or l- uh, let me rephrase that.
• 22:29
  It doesn't seem like it is something that should be so far outside the wheelhouse because it's 2026.
• 22:36
  It's America.
• 22:37
  Can't we do these things?
• 22:40
  Y- w- we supposedly, but h- how, how bad is it for him?
• 22:44
  Here, here's another, uh, ex kind of celeb that's, uh, I guess in our friend group on there, Kaya Rhodes.
• 22:51
  She's, her m- she's made a firm commitment- Textiles to rebuild textiles in this country.
• 22:55
  So like metals, where if you go back in some people's living memory, we were the dominant player globally.
• 23:01
  I think the stat is in 1950, America made 90% of the global textile production.
• 23:08
  So all the little mill towns, if you drive through Atlanta to Charlotte area, uh, and up into Virginia- Go up to Massachusetts.
• 23:14
  Yeah every little town is called Mill, right?
• 23:17
  Howell Mill and Aden's Mill and all these little mill towns.
• 23:21
  Well, it's'cause they used to have mills, and they were doing one of two things.
• 23:25
  They were either milling, you know, agricultural product, or they were milling, um, for the textile industry.
• 23:31
  I, I grew up just down the road from Hickory, North Carolina, and that used to be the furniture capital of, like, the world, right?
• 23:37
  Handmade, high-end furniture.
• 23:39
  We'd drive up there to get it at factory direct discount.
• 23:42
  Nobody in Hickory makes furniture anymore.
• 23:45
  It's not there anymore.
• 23:46
  And so we have, I call it selling our birthright.
• 23:50
  We've sold our birthright where we were once global leaders in an industry, and now fast-forward to today, the US makes 2%, 2% of the world's textiles.
• 23:59
  So we went from 90% all the way to the bottom at 2%, and here's Kaya trying to pull it out of the- Fortu- fortunately, we have Kaya who's gonna fix that.
• 24:09
  But, uh, I've been on the journey with her, uh, a little bit, and so what I've witnessed her do is she hangs out with a bunch of 70-year-olds Because those are the only people that know anything about textiles.
• 24:21
  That know it.
• 24:22
  Right?
• 24:22
  Yep.
• 24:23
  And the whole point she's trying to do is, hey, we have this agricultural opportunity.
• 24:27
  For example, in America we make a lot of wood, we make a lot of cotton, so we should be able to make advanced textiles in this country based off our own- Yep feed that we're already capable of growing with US-made farming.
• 24:39
  And, um, but the getting from the field to the store shelf or to the, the st- you know, the starship fabrics that are gonna be needed in, in a lunar environment is such a leap with all the intermediate processing and, and dyeing and weaving and, um, that happens in the midst before you get to a finished good.
• 25:00
  She just hangs out with 70-year-old people'cause they're the last living people that re- understand- That know it how to make fabrics.
• 25:07
  So- All right.
• 25:08
  So let's- That's where I see the- shift, shift for a second.
• 25:10
  Let, let's talk...
• 25:11
  So we've obviously got this huge knowledge gap.
• 25:14
  You can hang out with 70-year-olds.
• 25:15
  You can try and find some, you know, retired, uh, Soviet, you know, gurus if, uh, you know, maybe, maybe a couple folks still alive.
• 25:24
  Clearly best time to try and capture knowledge was 10 year, 10, 15 years ago.
• 25:28
  Second best time is now.
• 25:30
  That's right.
• 25:30
  We're talking through AI.
• 25:32
  I'm with you, you know, huge opportunity.
• 25:34
  Um, I'm skeptical on a lot of the applications, but one that I've seen they are very, um, ha- have seemed to be very effective with is that speech-to-text, um, capturing of it.
• 25:45
  Um, one of the things that, uh, I, I, I wanna hit on and get your thoughts is, you know, let's talk solutions.
• 25:51
  How can your shops, you know, whether we're talking, you know, small job shop type place, 50, 75 employee type place or something bigger than that.
• 26:01
  You know, obviously process, procedures, standards, all that's great, but again, doesn't happen overnight.
• 26:07
  What are some ways that you think- uh, you know, companies can start capturing that knowledge.
• 26:13
  I've got a couple ideas, but I wanna, wanna see what you've, your, your take is first.
• 26:17
  Yeah.
• 26:18
  My take would be, I'm about to explain the business plan of a AI integrator for you.
• 26:22
  So now that we have the frontier models and we have the industry niche models, right?
• 26:27
  And th- and thank you Silicon Valley for funding those, to get those built.
• 26:31
  What you can do with that replies, uh, you know, basically relies on you installing that at a company.
• 26:37
  It has to be installed.
• 26:38
  Uh, you see Palantir doing this great with a lot of big companies in the government.
• 26:43
  Yeah.
• 26:43
  But for smaller businesses, I don't, I don't know that they're gonna be able to go that route.
• 26:47
  So what you're gonna need is an AI integrator, which is somebody like you or me, or somebody in the audience that knows their way around a business, that has learned how to install AI systems at a company.
• 26:57
  And then why do they have to be installed?
• 26:59
  Can't you just buy ChatGPT?
• 27:01
  Can't you just buy, uh, you know, Claude or whatever?
• 27:04
  Well, the reason is, number one, you have no idea what the data privacy of those large company platforms is.
• 27:13
  You have no clue.
• 27:14
  Just like we've learned over the years- Yeah not to trust some of the big tech companies with our data.
• 27:18
  Maybe they'll do something in the profit motive with it.
• 27:21
  You have that same worry here, is that- It's a product form So you have to first understand what are, what are the data constraints and the data privacy.
• 27:29
  And the second thing you wanna do as you turn the AI up to be able to ha- do more things in the business, you need to have a firewall of your AI against your business.
• 27:38
  Because we have seen, especially with all this, uh, OpenClaw stuff, how the AI can just go rogue and we deleted all your emails.
• 27:45
  Just absolute nightmare.
• 27:46
  And then it apologizes to you later, so polite.
• 27:48
  "Hey, you deleted my data." My bad.
• 27:50
  "So sorry." Whoopsie.
• 27:52
  Uh, and so do you have the right backups on your data?
• 27:56
  You know, do you have a firewall in place?
• 27:59
  Just to share some secret sauce, what we do at Chang Robotics is we've set up a secure AWS cloud.
• 28:04
  We run our AI agents in the AWS cloud, totally isolated from our business operating environment.
• 28:11
  And that means it only knows and only does what we feed it, right?
• 28:14
  So a human is in the loop on the information that gets fed into the AI.
• 28:18
  And if it goes rogue in the AWS cloud, eh, it'll be unfortunate, but we're still- I- it's air gapped.
• 28:23
  It can't, can't get to any of that, that master data core record stuff.
• 28:27
  Yeah.
• 28:27
  And so you gotta have a little sense of policy.
• 28:29
  Like, most business owners probably don't know what, what the first policy they'd make is.
• 28:34
  How are we gonna operate with this?
• 28:35
  So it will operate very similar to your financial controls.
• 28:38
  In every business you have financial controls.
• 28:40
  How we invoice, how we do proposals or contracts, how we recognize cash, you know, those sort of things.
• 28:46
  So you're gonna have to have an AI policy that's, that's very similar in structure to your financial, uh, reporting.
• 28:54
  And, um, and so an integrator person or company needs to go in to most businesses and help them do that.
• 29:01
  The other thing is you don't wanna waste money, so you really need to identify what your priorities are.
• 29:05
  Is it we never wrote down our standard operating procedures, and we've got the top 10 guys are about to retire, so we need to get those recorded ASAP, right?
• 29:15
  So you gotta identify your priorities of what is it you're trying to use the AI to capture, and then help, hopefully help you organize so that you have some continuity in the business.
• 29:24
  So that would be my answer is every, every small-medium business either needs an internal champion that is the integrator, or they need to be actively looking for an external integrator that can come in and help them install the AI product across their business.
• 29:40
  I, I mean, all of that, uh, I would agree.
• 29:42
  I think that's, you know, ideal scenario.
• 29:44
  That's probably going to dramatically limit your exposure to chaos events, if you will.
• 29:52
  Um, I'm gonna take a slightly more, I guess, less polished approach to it and say in my mind, every shop that has, or every factory that has people on the floor doing things, strap a body cam to them Recording all day long everything they're doing and fee- take that basically that data, audio, video, feed that into your models and create a transcript effectively, and then you build off of that.
• 30:21
  So i- it's, uh, I'm taking a very crude description of it, of y- you take kind of synthesize real world what people are doing, try and translate that into a transcript or, or action of, of what's going on.
• 30:35
  That's gonna start to tell you some of the things that they're doing.
• 30:39
  But I think the real opportunity isn't so much what you're able to capture in that, it's what you don't capture.
• 30:44
  Or, or rather you see John did X thing.
• 30:49
  All right, you've got a whole separate workflow activity of, hey, let's follow up.
• 30:53
  Why did you make that decision?
• 30:55
  What, what made you decide to adjust it there?
• 30:57
  What- Mm-hmm you know, all the questions you have to ask because you can have a You can have somebody following around, you know, uh, y- your operators all day long, writing down everything, taking notes till they're blue in the face.
• 31:14
  But it- nine times out of 10, it isn't the things you observe, it's the things that are going through their head that they aren't...
• 31:23
  It, it's the decision-making and how they got to that decision.
• 31:27
  Right.
• 31:27
  The actions itself I don't think are the complicated or difficult parts.
• 31:30
  It's w- how you got to it.
• 31:32
  That's where that knowledge gap is.
• 31:34
  That's where the 30 years' worth of knowing what's wrong, what's right, how to fix it, that's what we have to capture.
• 31:41
  And you're not gonna get that with your video, you're not gonna get that with, you know, a questionnaire necessarily.
• 31:46
  You know, you've gotta dig into those things.
• 31:49
  Yeah.
• 31:49
  So I, I think the, just kind of playing back with the engineer brain of, of what you said, using AI in that way to begin to record the way you run your operation from the experts that do the operating will lead to AI-optimized compliance.
• 32:08
  'Cause now you'll actually have an audit trail, a track and trace if you will, traceability on what happened to a product from the time it came in till the time it went out the door, and who did it, right?
• 32:20
  And then you'll, you'll also, I think you'll probably be streamlined on enhancing your standard operating procedures, which to me is a good lead in to the, the Six Sigma philosophy.
• 32:32
  Right.
• 32:33
  Is that, you know, a- an AI-optimized physical workflow on the plant floor can, can now the AI can give you real-time feedback and do some of that Six Sigma work without a human trying to crunch everything and all the, the Kaizen type, you know, walkthroughs.
• 32:47
  It, that can happen somewhat passively.
• 32:50
  It's like,"Hey, did we, did we notice that we're spending eight minutes at this work cell for rework every time?" You know- You can monitor stuff and a lot- Yeah uh, w- without harassing everybody non-stop.
• 33:02
  I mean, and that's- But I would say I'm probably never gonna put cameras on my workforce, and it's just me being old-fashioned.
• 33:08
  I think it's creepy.
• 33:10
  Uh, I don't think most of my- Oh, I don't disagree at, at all.
• 33:12
  I, I, I would- I would- I think it's creepy as all he- a- as all hell.
• 33:16
  So what I would say is, um, is we already have plant monitoring.
• 33:22
  Every plant at some level has monitoring.
• 33:25
  It might not be what you want, but it's got something.
• 33:28
  Uh, so what might make sense is as, is almost a part of, you know, every...
• 33:33
  I'll do a buzzword.
• 33:34
  Every building has what's called a BMS, building management system, and the BMS usually controls your air conditioning and the lighting and, you know, the, you know, the basic infrastructure of the factory What might make sense is, is basically to do an extension of the BMS and get some AI vision cameras in the ceiling of a factory that are not monitoring...
• 33:57
  They're not recording, like, your face and what you're doing.
• 33:59
  What they're recording is the motion of an activity, which highlights that, okay- Face, transitions every, every fourth production run, we spent 10 minutes at the rework cell.
• 34:09
  That's now been flagged.
• 34:11
  Now you, as, as the expert or the owner, can go follow up, why is that happening?
• 34:15
  Guys, why are we doing that?
• 34:17
  And I, and that's where I think it, it, it's so interesting.
• 34:19
  I mean, I guess my expectation is not, hey, AI is going to answer all of these questions.
• 34:24
  Certainly not in the near term.
• 34:26
  I think what it does is it gives humans the opportunity to know what questions they've gotta ask more.
• 34:33
  And it's not AI suggesting the questions, it's these are things that happened.
• 34:37
  Okay, business owner, operations lead, whatever, you need to find out why.
• 34:41
  So you go ask questions.
• 34:43
  That, I, I think I've said it a million times, it's a force multiplier.
• 34:47
  Yeah.
• 34:47
  So rather than watching, you know, hours of film, it's gonna say,"Here's the things that happened.
• 34:54
  Okay, your job is now to go dig into it." And then, and then two, I think assigning cost is really hard in a factory.
• 35:01
  It's really hard to track cost of a factory because of all this touch and, and you, you don't actually, especially on the financial tape, you don't really know what's happening in the factory.
• 35:10
  It's, it's expert knowledge, tribal knowledge most of the time.
• 35:12
  So I think there's another interesting opportunity, which is, hey, last year our, as we were manufacturing this SKU or this product run, it was taking, you know, a total of 24 hours of processing time to factory.
• 35:25
  And then this year we did some product updates.
• 35:28
  Our customers made some custom requests.
• 35:29
  It's bumped up to 32 hours.
• 35:32
  But so we're ta- that same product is occupying 32 hours on the plant floor.
• 35:36
  We might not have known that, right?
• 35:38
  That might not have been obvious from the reports we get.
• 35:41
  But then now we can challenge the assumption, well, should we be charging more for that?
• 35:46
  Is there some first principles, uh, type engineering work that could just simplify the process?
• 35:51
  Why, why would- Do we need to go down the DFM rabbit hole with this?
• 35:54
  Is this really the best way to make it?
• 35:56
  That's right.
• 35:57
  Why, why did it suddenly bump up in utilization?
• 36:00
  And that is such...
• 36:02
  The way we view that, if y- and I know you've read plant P&Ls and so do I, it's such a lagging indicator because that shows up, like, at the end of a year.
• 36:10
  Maybe at the end of a quarter, but really it's the next year you realize,"Oh my gosh, throughput went down.
• 36:15
  What happened?" It, it, it's, it is one of the most valuable indicators, but it's not something that you're watching on your screen- That's right or you're not getting anything done.
• 36:24
  I, I, I, that's, I hadn't thought of it in that capacity, but I, it's, you bring in the vision component of it, I think there's huge upside there.
• 36:30
  I've got one final detail, uh, application for you.
• 36:33
  This was pre-AI.
• 36:34
  I think we deployed this in, um, 2021 We took a major beverage brand, so it's one of the top threes, so y- everybody would know who they are, and we took one of their distribution centers, and they were having a problem with truck driver overcharges.
• 36:50
  The reason the truck drivers were overcharging them, this is a distribution facility for beverage, so many trucks in and out per day- I know the time was because the, the, uh, fact- the warehouse team was either not unloading fast enough or not loading fast enough.
• 37:06
  So what we learned through that project was the truck, the truck driver, the trucking company gives you two hours for free.
• 37:12
  I'm gonna pull up to your dock, and included in the charge of the haul is you get two hours.
• 37:16
  These guys were routinely hitting eight hours to unload or load a truck.
• 37:20
  So not only are they killing the truck driver's day, but they're also killing their own cost because they're paying overages.
• 37:27
  So what we were able to do for a very low cost was we, we, I'll say rigged, um, we assembled a IoT monitoring system where we could monitor the duration of trucks at the dock.
• 37:39
  That was data they didn't have, and so we just put little sensors, occupancy sensors.
• 37:43
  When the truck pulled in, bing, truck's there, then we just measured the time.
• 37:46
  Like the, uh, the parking lights at the garages.
• 37:48
  Yep.
• 37:49
  And we tied that into a clock inside above the loading docks.
• 37:53
  Everyone got a clock, like the CrossFit clocks, you know- Yep, mm-hmm when they set the time for the workout of the day, and it would start the timer And- I love it you had to have the truck either unloaded or loaded in an hour 45 to give yourself cushion on the two hours.
• 38:08
  That simple transparency increased the performance and the prioritization of the workforce inside the plant,'cause now they knew,"Oh, this truck, this truck's been waiting for an hour.
• 38:18
  We haven't even touched it.
• 38:18
  We better get on it,'cause it's gonna take us 45 minutes to unload." Um, the other thing that we did was we tied it into a little app so that people could look at it remotely and see was it moving the needle, give them a little historian and graph.
• 38:32
  I do wish AI was around,'cause we had to code that app and it took a while, versus now I could vibe code it in 30 minutes.
• 38:37
  Vibe code it, yeah.
• 38:39
  Um, but yeah, that, that's an example of little amounts of information dramatically increase the performance of a business.
• 38:47
  And one of the reasons they didn't perform better prior to us doing this project was they, they didn't know how to measure it.
• 38:54
  Right?
• 38:55
  They didn't know what- Uh, on a, uh, on a similar note, one engagement I was on in consulting world was rather than trucks, we were tracking basically component movements throughout the factory.
• 39:05
  So it starts one place, all the different stations it has to go to.
• 39:09
  And we used, uh, wearables, again, the IoT sensor thing basically, to figure out where are these dots moving for part.
• 39:16
  You know, a- and you can create your little...
• 39:19
  It, it's a flow chart, if you will, but it was a visual representation looking at the factory floor, all different machine stations, idling, all those pieces of it.
• 39:29
  And based on that input, we were able to take and say,"Okay, here's how we can move a couple of these machines around so, A, it's gonna cut out 75% of just the distance and time that you're moving one to another." Transport.
• 39:42
  And paying, paying people to carry stuff, yep.
• 39:44
  Yep.
• 39:45
  They didn't go so f- you know, it, it wasn't...
• 39:48
  The ultimate solution was not the, uh, the sexiest, if you will, but it ended up where basically ended up with painted lines on the ground for different...
• 39:57
  After moving machines, you have different sequences, different activities follow different color painted lines.
• 40:02
  It's if you are doing one particular thing, you are on yellow line, you are on red, green, blue, whatever,'cause it's, that is your happy path.
• 40:10
  That is your best way to go to it, and it comes down to- Machines had been moved in there because as they grew, as they needed more capacity, they put them where they could.
• 40:19
  And you have to, you know, eventually look and say,"What's the optimal path to go to?
• 40:24
  What is A to B to C instead of A, D, F all over?" You know, it's, it's a change.
• 40:30
  People hate change.
• 40:31
  Yeah.
• 40:31
  But people have also very quickly like change once they realize, oh, wow, this is a lot easier.
• 40:37
  I'm not running around, I'm not dealing with all this.
• 40:39
  So information, you know, I, I think comes back to having that information.
• 40:44
  You know, you're talking about trucks, having that insight into what's the idle time- Yep for load and unload.
• 40:49
  What, what's the, you know, part movement throughout the factory?
• 40:52
  A- and by the way, they have this at ports.
• 40:54
  Yep.
• 40:54
  So a, a fun fact for you, we've been, we'll reveal it soon, but we've been digging deep into the port infrastructure, uh, of our country.
• 41:03
  Fun fact, most ports are operating at 50% capacity because of the crane.
• 41:09
  There's not enough staging area behind the crane to unload a ship, so they have to take breaks while they clear out containers that just got unloaded so they can resume unloading.
• 41:19
  So as a nation, you know, I'll generalize this, somebody can challenge me on the precise percentage, but we have a 2X latent capacity in our ports because we don't have the ability to manage the things that just got offloaded right behind the crane.
• 41:34
  And so that causes a little bit- Idle time those little minutes, you know, minutes turn into hours, hours turn into days, and over the course of a year you're losing almost half the year on crane time.
• 41:44
  So, but, uh, back to your, uh, your, um, story about optimizing the factory.
• 41:49
  I have two observations before we wrap.
• 41:51
  Number one, that what you described, we call it in the industry, we call it a spaghetti diagram,'cause I'm sure you noticed your parts- Yeah are just bouncing all over the place.
• 42:00
  But then you put another part on there and it's bouncing all over the place.
• 42:02
  You do another, and before you're done, your chart looks like a plate of spaghetti, but you just got all these lines moving.
• 42:07
  You overlay them all.
• 42:08
  Yeah.
• 42:08
  You got turns, and whips, and all that good stuff.
• 42:11
  Yeah.
• 42:12
  And then- Throw a couple meatballs in there and you're good yeah, so spaghetti diagram, that's free for the guys that are new to the, the industry.
• 42:18
  Uh, the other one is a consultant joke.
• 42:20
  So they hired a consultant and all they got was painted lines.
• 42:24
  Exactly.
• 42:25
  No, they got some cool cameras out of it.
• 42:27
  They, they got some, you know, nice SOWs and...
• 42:30
  No, they got a bit more than that.
• 42:32
  No, I'm kidding.
• 42:32
  But yeah, basically you're, you're not wrong.
• 42:34
  But again, you know, it comes down to y- the, the information knowledge.
• 42:39
  You sometimes need fresh eyes.
• 42:40
  I have comments I'm not gonna make on the consulting thing- Yeah, yeah having been, uh, retired from that.
• 42:46
  But, um, I mean, to your point- Let's not do this live it, it, it, it takes, it, it comes down to if you don't have information, you can't measure things.
• 42:56
  If you can't measure things, you can't improve them.
• 42:59
  Yes, sir.
• 43:01
  All righty.
• 43:01
  All right, yep.
• 43:03
  Great episode today.
• 43:04
  We'll, uh, keep digging into the workforce'cause that seems to be our, uh, our, our sweet spot.
• 43:09
  But to everybody listening, thank you for tuning in to another episode of the Reforge Podcast, whether we're talking robots on the factory floor, policy that moves the needle, or the next generation of makers, our goal is always the same: to prove that reindustrialization in America isn't just possible, it's happening right now, and it's pivotable.
• 43:26
  If you enjoyed this episode, make sure you subscribe and follow the Reforge Podcast everywhere, and check out thereforgepodcast.com for more.
• 43:33
  Until next time, keep forging the future Thanks for tuning in to the Reforged podcast.
• 43:44
  Make sure to like, subscribe and follow us on all of our social media@thereforgedpod.
• 43:49
  We will see you next time