Episode #81: The Missing Layer in Reshoring: Andy Jenks on AI Diffusion in Manufacturing
In this episode of U.S. Manufacturing Today (powered by Veryable), host Matt Horine talks with Andy Jenks, co-founder and CEO of Jakib, about the overlooked software layer behind reshoring and why it’s deciding which manufacturers scale and which stall. Jenks traces his path from Alexa Internet and building Amazon’s first recommendation engine to Drive Capital in Columbus, where a 2021 demo of an early large language model changed how he saw AI. He explains the “diffusion gap”: buying AI licenses isn’t the same as putting AI to work, and real adoption usually starts with basic engineering that gets disconnected systems talking to each other. His view is that AI has started a build-software revolution, not a buy-software one. Jenks describes how Jakib embeds engineers to shadow employees, ships production software within weeks, and then hands it off while helping clients hire their own teams. Examples include a 3PL that grew revenue without adding headcount, a U.S. wireless module maker bringing chip design work home from China, and AI-enabled walkie-talkies on a factory floor where phones are banned.
Links
- Andy on LinkedIn
- Jakib AI
- Matt Horine on LinkedIn
- Veryable Is Revitalizing U.S. Manufacturing
- Sign Up on the Veryable Platform
Timestamps
- 00:00 Welcome to the Podcast
- 00:42 Software Behind Reshoring
- 01:51 Andy Jenks Origin Story
- 04:31 Seeing AI Early
- 09:57 The Diffusion Gap
- 13:51 Build Not Buy
- 17:52 Jakib Accidental Start
- 20:54 Abundance Over Cost Cuts
- 23:44 Deployed Engineers Playbook
- 28:01 Reshoring Chip Modules
- 33:06 Factory Floor AI Example
- 35:39 Monday Morning Plan
- 38:11 Wrap Up and Resources
Episode Transcript
Matt Horine: [00:00:00] Welcome back to U.S. Manufacturing Today, the podcast powered by Veryable, where we talk with the leaders, innovators, and change makers shaping the future of American industry.
Welcome to US Manufacturing Today, the podcast powered by Veryable, where we talk with the leaders, innovators, and change makers shaping the future of American industry, along with providing regular updates on the state of manufacturing, the changing policy landscape, and more. Today, we're talking about the part of the reshoring effort that never makes the ribbon-cutting photo, which is software.
We spend a lot of time on this show talking about CapEx, supply chain, labor, but there's a layer underneath all of it that most small and mid-sized manufacturers have never had to build themselves, and right now it's deciding who scales and who [00:01:00] stalls. My guest today is Andy Jenks, co-founder and CEO of Jakib.
Andy started as a software engineer in San Francisco, founded several companies, and then spent years as a VC in, in uh, based in Ohio and investing in data and compute infrastructure that became the AI boom. Today, his engineers embed inside founder-owned freight, construction, and manufacturing companies and build the software that runs the work.
And then, most importantly, they hand it over. We'll get into why the mid-market is stuck on AI, why Andy turns away customers who just want to cut costs, and a, Ohio manufacturing story that went from zero to a real business in about a year by bringing a product home from China. Andy, welcome to US Manufacturing Today.
Andy: Hey, thanks for having me, Matt. Appreciate
Matt: Yeah, we're very excited to have you on the show today. This is just a continuously recurring topic that we hear all about this, this type of layer within the manufacturing world, and I think you can provide a ton of insight to our listeners. So with that, we'll jump right into it and maybe get started with a little bit about your background.
You started as a software engineer in San Francisco. You built several [00:02:00] companies and then did something a lot of people out there swear they'll never do. They moved back home or moved to a part of the country they didn't think was necessarily where they would be. Take us through that path a little bit and how you landed back in Ohio.
Andy: Yeah, I um, so I moved, I moved to California, software engineer in the late '90s. So showed up in 1998 and knew how to write HTML, which I had just learned about in college, and CGI if anybody remembers that from way back in the day, and learned how to write code that basically powered the internet.
Started my career at a company called Alexa Internet. Alexa was crawling and and basically indexing the web. This is pre-Google, or right around the same time actually that Google started. We were fighting Google over time for who had the largest web crawl. Alexa was winning most of those, and then Google took us and then, the rest is history.
Alexa, however, was acquired by Amazon, which is where I really learned what scale meant. Built the first version of the recommendation engine for the Amazon website. Again, this is-- we're talking year 2000. And then went on to start several [00:03:00] companies in San Francisco Bay Area. I was, founder, CEO VP of engineering, chief sales officer, like you name it, did it all.
And in 2012, was approached by some good friends who said, we're about to start a new venture firm. We're leaving Sequoia, and we're gonna start Drive Capital. What do you think? Do you wanna join us?" And and I had always been an investor myself angel investing at the time in 2006 when I started.
They called it friends and fools as opposed to angel. Hopefully I was a friend. Sometimes I feel like I was a fool. But but really wanted to get into the investor side of the house. I felt like I had a lot to offer to young entrepreneurs as they were starting their careers. The, the big difference was that the firm was gonna be started in Columbus, Ohio not San Francisco, and that was that was something that was, that I was not expecting.
And it hadn't really been done. It has been done, but it hadn't really been done at the scale or the size that Mark and Chris wanted to do. So packed up the bags [00:04:00] in 2015. It took me three years to, to get the the kind of confidence to talk to my wife and my kids about leaving San Francisco.
And 2015 started as a general partner in Fund One, and was with Drive through through Fund Four.
Matt: That's a, it's a great story kind of, going back that far and definitely we'll take some tips from you maybe after the show on talking about transfers or moving the family. We just went through one ourselves so we did that whole we're moving to somewhere we never thought we'd move to, so that's always a-- I'm glad it's a universal experience.
At Drive you were investing in what we now call AI before anybody really called it that. You've said one seed deal in 2021, around 2021 flipped your thinking on it. What did you see under the hood that maybe others were missing?
Andy: Yeah, I was, I was always kind of deep nerd. We were building infrastructure, like that was the, the... Those were the companies that I had started in the Bay Area. They were, I would call them deep infrastructure companies with kind of a UI on top. What we saw in 2021, 2022 [00:05:00] timeframe was something that was totally different than what I'd seen in the past, and that was a company called Cohere.
They were part of the team had left Google Brain. One of the authors of the paper, "Attention Is All You Need," was one of the co-founders, and they really opened my eyes, so to speak, to this world of big data, which is what we were all messing with at the time. If you remember, there were the NoSQL databases, there was the Hadoop ecosystem.
They called it big data, and there was gonna be this big re-platform. And we heard at the time what we're hearing a little bit today, which is there's a lot of investment without a lot of return. This, however, was something totally different. When I first saw an LLM in action in, in 2021 in Toronto of all places not in the Bay Area at the time it really kinda changed my mindset and my thinking as to what's possible.
It had terrible hallucinations. So to be clear, at that time, point in time, you'd say, "Who's Andy Jenks?" And it would make stuff up. It would say that I was, an Olympic power lifter and y- I I was a [00:06:00] movie producer, of which there is one called Andrew Jenks but that's not me. It kinda pulled some information about me, and it made up this really great story actually about me that I liked, but it wasn't accurate.
But you could see the genesis for what would become, the, the GPT landscape that we see today.
Matt: Yeah, there's-- I, I do remember back to that 2021, '22 timeframe. We joke about the AI from the consumer standpoint about college kids using it on their homework, and that was what a lot of us had a few years later. It's like, "Am I cheating?" It's like, you are not cheating by using you know, these models and all those things.
But those early days were very interesting. Being early in this, investors usually make their money on the picks and shovels in a tech revolution. Why did this one, why did this one feel so different to you?
Andy: Number one, the picks and the shovels were are where if you look where the returns are today SpaceX, Anthropic, OpenAI, those three IPOs will generate more wealth in the Bay Area than the previous, what is it, 40 years or something like that of venture [00:07:00] combined. So there is money to be made in the picks and shovels for sure.
It was just, I was shocked at the time when we were talking to the Cohere team, they were raising money in their series A, so they had done a series C. Their series A round was not necessarily to hire people, it was to perform their training runs. They needed GPUs, they needed inference, they needed all of the infrastructure that was going to help them do...
I, and I'm trying to remember exactly, but I think it was like a 5 or maybe it was a 20 billion kind of parameter model, which today is, what I have running on my laptop. But the, the point is at the time to train that it was gonna take three months, it was gonna cost $25 million, it's gonna be very expensive.
And so instead of raising money for people, which is traditionally where as an investor I would invest, and those first dollars would go to making sure that you build the best world-class team you possibly can. These guys already had a world-class team. They didn't necessarily need to augment that with additional human capital.
They needed capital to be able to do their training run. That to me was like a big aha [00:08:00] moment, which is, wow, this is gonna be expe- it's a great place to invest, number one if you can find the right company. But number two, you need a, a enormous amount of capital to be able to back these companies long term.
And if your series A is gonna be, 40 million, 50 million 60 million, we've seen it play out over time. Your round sizes are gonna get to the billions pretty quickly. And that to me was the kind of awakening moment, which is I want to invest here. We had a big fund, like Drive is a $2.2 billion AUM fund.
It's a big enough fund to be able to back those early kind of investment rounds. But at the same time, it was like, wow, you really need, multiple billions of dollars to be able to invest to protect your ownership over time. So where other pockets could we start investing in? Again, maybe some picks and shovels here and there, but at the same time, we didn't know what it was gonna look like at the end of the day.
There were still a lot of hallucinations. The technology was not where it is today. Is this a flash in the pan? There's a really funny quote [00:09:00] I'm trying to remember who it's from. It's, it may be from Venky. Can... I don't know exactly who it's from, but they're like, a great place to lose credibility as a venture investor for, 20-plus years was artificial intelligence," right?
And so it was early enough where you still h- had a little bit of that sentiment in your brain. And so we made a ton of investments. We made great investments. We invested in a company called Greptile. We invested in, a legal company. We invested in several great businesses, Vox OD1, like great businesses in and around picks and shovels.
But at the same time, like the landscape was very different. The dynamics for making those investments was different. The capital was different. Things were just different enough for you to say, "Wait, hold on let's take a pause. Let's look at the landscape and kind of reevaluate." 'Cause you have a responsibility to your LPs, right?
You have a responsibility to make sure that capital is going after great opportunities.
Matt: Yeah, it's-- that brings up a lot of interesting points about, the timeframe and the uniqueness of it. I think, when we first spoke we [00:10:00] talked about what you framed as the diffusion gap. So this is a little bit of a segue and something at the root of what you do. Something-- we-- I looked over data in preparation for this show and just saw some things that didn't add up.
I think Rootstock Software's 2026 State of Manufacturing Technology survey found something around 94% of manufacturers say they're using some form of AI, but there's other data on mid-market companies, very recent publishings, found that 73% are still in the testing phase, and they're not, it's not a single company surveyed had reached full company-wide deployment.
You called Jakib a diffusion company. What does that mean, and why aim it squarely at businesses doing, a certain amount of revenue or a certain segmentation of the market?
Andy: Yeah, this is exactly what we... The conclusion we came to, aft- after I left Fund IV, the conclusion was we need to understand how companies are going to be using this technology before we start continuing to invest dollars if we are not going after the picks and shovels, the traditional kind of layer of the [00:11:00] stack that early investors were going after.
And so when I left Drive, it was really on this kind of quest to find the use cases inside of the businesses that were leveraging AI. This is where that diffusion comes in, right? Which is you show up to a company, they say, "We've bought, a license for ChatGPT to the, the sales team, the marketing team, the this team, the that team."
That's not diffusion. That's not using AI. That's using a, a small piece of the what's possible. And so what Jakib does is Jakib actually shows up, we interview, we look for outcomes, and we are truly bringing the power of the frontier models and some just regular old good engineering to these businesses.
And so it's not surprising that you hear stats like that, where if people say the large companies might say "We're using AI." How are you using it?" The answer is, "We're using it because everybody has a license to Claude." That's not really reusing AI, right? Maybe there is a small sliver inside the organization that's actually using Claude code.
They're doing the things, with the model [00:12:00] that are scratching the surface of what's possible. And it's also not surprising that the mid-market is like, "Man, I don't even know. At some point, somebody's gonna bui- build a product. I'm just gonna buy it, and as soon as that product's available, I'll buy it, I'll implement it, and then I'll, quote, 'have AI.'"
That-- They're wrong.
Matt: Right?
Andy: This is why it's different, is we are in a build software revolution, not a buy software revolution, and that's the big insight that Jakib has really, taught me, quite frankly, which is we are building software in a way that could never be sustainable or possible previously without the cost curves of taking, really difficult C++, really difficult Perl, really gr- difficult JavaScript, all these different languages, and allowing a team like Jakib to be able to build that software better, faster, cheaper at production.
So these are production systems with all of the harnesses and all of the security [00:13:00] and all of the backup and all of the enterprise kind of things baked into it in a bespoke way for these mid-market companies. And that is the big revolution of AI, in my opinion. What we always say is people say, "Hey, can you bring AI to my business?"
It's like, "Sure, but let's start with twenty sixteen engineering." Twenty sixteen engineering says, system A doesn't talk to system B, and there's a spreadsheet, and there's a Goo-Google Drive, and there's a this and a that and a whatever. Our team uses the own-- our own harnesses, our own tools that we've built that are, enabling us to move a lot faster, hooking all that stuff together.
Once that stuff gets hooked together, then you can bring the magic box and you can say, "Let's run, a deep, a deep research against, this report or this board material or, this strategy that we have and use all of our data." That's where you get the magic, but it starts with good old-fashioned engineering.
Matt: Yeah. S- speaking of that good old-fashioned engineering, you made a really great point. And this is part of that, build versus buy conversation, but when we first met, and this is an [00:14:00] observation for a lot of manufacturers it's something that I see with manufacturers as well. Those companies usually have great IT people or what we would call the classic, IT people.
They bought Workday, they bought the ERP system, but they've never really built something from scratch. Is that... what's the real barrier there? Is it more than budget, or is it a willingness, or what do you think the real barrier is to that? They're just not a software company?
Andy: Yeah I think that, there's a famous saying by Marc Andreessen that everybody's a software company, software will eat the world, all of that stuff. We have run across amazing IT teams and a traditional IT team, you've got a leader who understands how to procure, buy, implement, roll out software platforms to their kind of customer, their customer being the internal consumer of that software inside of whatever organization.
We've run across great people that know how to do that. The building of software from a GitHub repository effectively, and being able to tie all this stuff together, software engineering is not IT. And it does not mean that one is better than the other. It just [00:15:00] means that they need to work together in unison.
And so what Jakib brings to the table is we bring the software engineering expertise. We work with the internal IT team who have set up the systems and the platforms and the process. We bring the software engineering knowhow from an amazing team of ex-founders, public company CTOs, like you name it, we've got a great team.
And we show up and we put these two things together. The other thing that I believe to be true that has historically been the pattern over time inside of these organizations is when IT shows up and they say, "Hey, it's time. We need a new email system. We need a new ERP. We need a new CRM. We need a new whatever database back," they sometimes, depending on the size, will hire a consultant to go tell them how to do it, tell them what to do.
If they don't have a consultant, doesn't matter. They'll then run a bake-off. They'll then evaluate a software platform based on a whole set of criteria. They'll run them through the security audits, and then they'll pick one. That process takes 12 to 18 months to get one picked, [00:16:00] stood up, starting the migration path.
Sometimes that migration can take two-plus years. It just depends, again, on the size. And that historically has been a really great muscle that IT has built over time. And I would say that this is where, again, this is no, throwing shade at IT, but this is where IT needs to have the moment that I had in 2021, which is this is totally different, in that if you start down that path of picking software that way in the world of AI, you will be too late.
It will be too late for you. The world of procurement and, quote, "buying AI" is not showing up and saying, "Okay, OpenAI, okay, Anthropic, come sell me AI." It is truly, let's figure out which outcomes we're trying to get to. How do we build internal software systems and platforms to go after those outcomes? And then look at the underlying platforms that support those outcomes.
What we do is people say are you gonna throw out..." Yeah. "Should we move things to Snowflake?" It's like I don't know. Today we're on, you know- This old version of Oracle. It's like, great, stay there. Your business isn't [00:17:00] broken, and by the way, your IT n- team knows how to run it. They know how to maintain it.
They've been... M- maybe they hate it, but they've been doing that for years, and your business is not broken. Leave that stuff in place. Let us go, and over time, start stripping away the features that you don't need left in the legacy platforms. Move them over to a more modern platform, and after 12 months, 18, 24 months, however long it is, all of a sudden you're like, "Whoa, we actually don't need this anymore because we've migrated."
And you can only do that as you're building software and your flywheel is very tight. And so that's one big thing that I think IT needs to start embracing and start thinking about.
Matt: Yeah, really said on the, the differences there, and I think we're gonna get into a little bit more of that on the show on the difference between, the, the classic software company some of those migrations. I'm sure, folks who are listening have certainly done an ERP rollout or they've done some kind of consolidation we don't wanna trigger anybody there.
But just before we jump into that j- a little bit more background on the company. Jakib started it a- started almost by accident is if [00:18:00] I remember you telling me correctly. You went looking for companies already using AI and, what else did you find that was like this is an accidental, like you've discovered something ver- very unique here?
Andy: It was the... it was accidental. Like I said, like, when I left, it was like: if we're gonna invest, we gotta figure out how they wanna use it. So that was the first kind of question. With the f- actually, the original thesis was let's start a fund to go find the great companies that are gonna be selling software to these other companies and blah, blah, blah, and then the rest is kind of history because we've pivoted that completely, and now we are Jakib selling them so- selling them engineers, basically four deployed engineers to go build software.
The other piece of accidental kind of knowledge that we pulled out, and I wouldn't call it accidental, I would just call it more it's not talked about, it's below the surface. These businesses that we work with, these are mid-market businesses, $100 million plus in revenue. It doesn't have to be, it's not a hard number, but that's where the that's where the, the scale really helps is like, 'cause you can move the needle if you've got 100 million [00:19:00] and you improve something by, you know, 15%, like that's a real, that's a real number.
But these businesses fall into two camps. They fall into the one camp of, "This is how we've always done it. This is what I wanna do. Can you please bring AI into our business so that way I can cut costs?" So that's one. The other camp, which is the more exciting camp and the one we work with, is we have 100-plus million dollars in revenue, and we see AI as this opportunity for us to leverage to go take market.
We see this as an opportunity for, and a lot of the businesses we work with, manufacturing, like you talk- this is the whole podcast, but manufacturing, construction, logistics, 3PL, 4PL, staffing. These companies are often multi-generational, and sometimes they are, sometimes they're not, sometimes they're first generation, but most of the times they're multi-generational.
And you'll run across people who wanna run the family business and just do same old, strip some costs out, and just do what they wanna do. We don't work with them because there's nothing wrong with that, but that's not the best way to use AI. The [00:20:00] best... I- if you're gonna do that, just buy technology and keep doing what you're doing, but most likely you'll be beat by the companies that are leveraging it.
And so we look for those companies that are leveraging AI. And what you find is when you start putting the what's possible in front of people, their eyes light up, they get really excited. And people do want to grow, they want to take market, they want to compete, they want to do that American Dream kind of thing, which is "If I work really hard and I put the right tools in place, I can go take market share."
And I feel like there's a second life almost to some of these businesses where when we show up, we start showing them what's possible, and then the entire business just starts getting better. Not because of us, but because we're tapping into something that they've always wanted to do and they just hadn't had the tools to do it.
Matt: Yeah, it's unlocking the dream, right? A lot of, especially like those family-owned businesses, the next generation does wanna do something, bigger or different and that's kind of the natural trajectory of things if you will. I think you and I talked about it, the abundance outlook versus this kind of this cost-cutting measure.
Just [00:21:00] out of curiosity, because I think there's a lot of this is entering the discourse and more in that framing lately, the abund- we can talk about it with data centers, we can talk about with any... no, particular stance on it, but, what does an abundance project look like next to a cost-cutting one as an example of AI making the business bigger instead of, just managing these systems or the software seats type structure?
Andy: Yeah it's it's a really good point to make, which is the, the cost-cutting is the p- when we say, "What are the outcomes that you're looking for?" And they say, "I need to reduce my back office." I've spent... When we first showed up to one of our 3PL customers, "We're spending $3 million a year with a call center, and I don't wanna spend that much money with a call center.
We need to save that money. We need to change it." It's like, "Okay, why are you spending that much money with a call center? Oh, they're doing a critical component to your business? I can show up and sell you 11 labs with a voice agent, with a rapper that does a bunch of stuff for you to save some of that money.
But it's so critical to your business. Why would you care about saving money if it's so critical to your [00:22:00] business?" The outcome was save me money. Where that went fr- to is when we started doing the kind of five whys exercise, which was why do you wanna do that?" We wanna do that because the only way for us to scale, the only way for us to grow, is to add more people to the call center, and we wanna be able to grow the revenue without adding a lot of people.
It's like, okay, that's a different conversation. Let's have that conversation where why do we need to add more people every time we add a new customer or a, a new route or a new carrier network or something like that. And the answer came out of a lot of meetings and a lot of questions and a lot of software development.
And then, by the way, over time, what started out with a, "Let's save money on this call center and replace it for me" was we actually replaced it, in month eight because we had to do a whole bunch of other things to start. And over that time, we continued to grow top-line revenue, grow gross profit, grow bottom line without growing headcount.
And as soon as we were able to prove ourselves [00:23:00] that, we were able to start divesting from the big expense of the call center. But guess what we did? We kept the call center as well for certain aspects of things that can't be automated, right? And so it starts with save me money. That's one side. That's not very interesting because the outcome is save me money, not interesting.
But then when you move to outcome, I wanna grow, and the only way I can grow is by adding headcount because this call center helps me, with my track and trace, it helps me with my delivery schedule, it helps me with all these things. It's like, okay let's tackle that problem to help you grow.
When you tackle the problem to grow, you by, kinda default, you start being able to trim other areas, and guess what? Costs comes down.
Matt: Yeah. Cost cutting is a kind of the natural progression of better efficiency, not, the cost out and the s- the stripping away. Just a couple other important points on the build it, hand it off, and walk away mentality and structure of this. Most software companies wanna charge you per seat forever.
I... Those, those numbers. I've been in the business. I know what it looks like. I I've been in some really, in the healthcare world, I was in a [00:24:00] really detested segment of the healthcare world that involved credentialing, and just the seat cost kept going up, nothing changed, no more value provided, just the same thing over and over.
You build the tool, hand it off, and help the customer hire their own engineering team. But the first step of that is you put those four deployed engineers. What does it look like when they show up, and what are they doing kind of the first few weeks? I know you just gave that example, but some of the step-by-step if someone was thinking about something like this.
Andy: Yeah. Yeah. And I love this part about our business, which is not trying to lock people in to seat licenses.
I'm really glad we're not doing
Matt: Yeah, we're, you, guys like you and I are we're free from the shackles of having to do that in the past in our work, right?
Andy: Exactly. The way it works is we kinda show up, we sit down with... It's often the CEO because this is often driven by, the executive leadership team saying we need to bring technology to bear on a certain amount of problems." We scope out the outcomes. There's typically two or three outcomes, and then we agree to start.
The start is the most interesting piece, which is we literally just shadow people. [00:25:00] We have engineers, like, these are, like, really high-quality engineers that could be writing code, but they're watching people work. They're literally saying "What are you doing every day?" "What's the worst part of your day?"
That question, "What's the worst part of your day?" often has so much insight underneath. And we start with the people of, oh, it's, I need my accounts payable is a big headache for me because, we have all of these lines of credit, we have all this debt. It has to flow, and so the faster we get paid, the more money we can pay down on this, and we can do that.
These are often, even though they're big businesses, they're often run on bank lines or personal guarantees or if it's a family-run business. And helping with things like that and when you get into the accounting process, don't switch the accounting pro- like, don't switch any of that stuff.
Just switch the process underneath of it. Make it more automated. And y- you'll hear that, Bill or Sally or whomever, they have to run a report before they leave every day, and they can't get home, and they wanna get home, and their mind is somewhere else, and they're gonna miss a soccer game if they don't run this report.
The report takes an hour and a half at the end of the day after they've had a [00:26:00] really long day, and that report sometimes is wrong. And they can't have it be wrong because it gets filed to the bank or it goes where it needs to go. Our team will watch how this all plays out, and then we start attacking those problems.
After we're done attacking those problems, we have what's called points on the board, where we show up. We just wanna put points on the board from day one. Typically within two weeks, we have a full bit of software that's production-grade software that is in production that is doing that piece of work for that person, and it's making their life better, and we go from there.
And then we just start tackling little components. It's true for deployed engineers, where we don't necessarily need to be in your office every day, but we are working with you every day. We are a SWAT team is kinda how we think about it. We show up. We do the hard problems. We do the sticky things. We make your life better with software.
Some of it, like I said, is 2016. Some is 2026. It just depends on what we need. And then over time, after 30 days, 45 days, you're given this update at the C-level or at the [00:27:00] executive level, and everybody's like, "Wow I've made so much progress," right? So it's not a per seat. It's not... we're literally building software, and then to your point earlier, it's like, when we're done building, we hand it over.
We help you hire the software team that's gonna take this over. We work with your internal IT team. If you have a CTO or a CIO, we help them understand how to frame up what does a software organization look like in this new world. Doesn't happen overnight. It definitely goes- For, oftentimes we say give us three months.
Three months often turns into six or nine months. But but our goal is to get diffusion into the US GDP, and that starts with the hardest of markets, and this is construction, manufacturing, 3PL, 4PL. These are the things that move America, and we really wanna bend that curve and make the productivity stand out.
Matt: Yeah, productivity and utilization is really the big story on reindustrialization.
To, to your point, I think that's unlocking a lot of, how do we unlock the capacity or how do we rebuild the [00:28:00] capacity, really important points. It's kinda transitioning a little bit here to the real world reindustrialization scope of this and where this actually affects manufacturers and people, and I think you've really hit a lot of the high notes on the abundance framework.
A, a lot of bubble talk right now, people comparing all of it to, dot com era, kind of the irrational exuberance. But if it's done the right way it funds the gaps a small or midsize manufacturer runs into every day, and you're using your own data, building your own solution, and feeding the business instead of, blaming whatever the headwind is of the week, which is common in manufacturing 'cause they're real.
It's, they're real headwinds and they're real challenges. But we talk a lot about reshoring and the process on this show like it's a building or it's a ribbon-cutting ceremony. There's a lot more to it than that, and some of the hardest pieces to bring home are the, the smaller pieces, the components inside everything we buy that nobody really thinks about until they can't get it, and I think supply chain told that story over the past half decade or so, and now people are finding out how hard it is to build things here in this world.
You had a customer you know that recently they were solving a big problem in this components landscape [00:29:00] and why couldn't the US just do this already? Where did Jakib fit in that and what was a little bit about that customer that was so critical?
Andy: Yeah, this i- this is this was really fun for the engineering team that was working on it because it, it is tech, but not tech. It's
tech. It's like the tech universe, but it's old school Silicon Valley tech. As we know a lot of chip manufacturing has moved overseas. The design of those chips is now all...
It mostly lives, like 90, I think 90% of it lives inside of Asia. And we're seeing it. We're seeing it play out day in and day out. And so this was a manufacturer of of modules, Wi-Fi modules or wireless modules I should say. So 4G, 5G Wi-Fi Bluetooth, these are the modules that connect things.
These are IoT modules. And Qualcomm's got the chipset, but you can't just take the chipset and just paste it on something and it works. You gotta actually put it in a bo- system on a chip basically that allows you to do things. Geotab, if [00:30:00] everybody knows who Geotab is, they are, the tracker of most freight in the United States, like those Geotab boxes.
They are on, A lot of the, semis that, that are driving around to keep track of where things are, right? And that's how a lot of these IoTs... so think about things like that. This company was basically started just over a year ago with no no customers at the time and a lot of expertise around the table, but they were licensing a lot of the technology from Asia to re-onshore a lot of the chips to get them manufactured in the United States.
There's, tensions between the United States and China. Who knows what's gonna happen with Taiwan? This is not a political conversation, it's just fact. And these entrepreneurs saw an opportunity to do that. They ended up finding the right partner. They ended up starting to bring a lot of that manufacturing back to the United States.
They started doing the manufacturing piece, not the design piece back here in the US. They got a lot of customers. They're now just north of $100 million. They're growing now. It's a real business that's manufacturing [00:31:00] these wi-wireless modules. The big component of it that was so hard to do, it's not can you buy machines from Germany and all these different places to put it together?
It's the actual IP that sits in underneath those, those chipsets to be able to put them all together so you can paste them on a board, so people can program them and do with them what they want. That design work for the chip and then the firmware and all the stuff that goes along with it, that all lives still in China.
And so there's the United States there's a US basically code, the FCC just ruled on it, and there's a connected car from a, having EV, Stellantis, Tesla, like you name it, like all these cars are connected now. There's a connected car rule that says anything after 2028 cannot have a China-designed and manufactured chip on it.
So what do we do? This company is the one that's gonna solve that. We, Jakib, showed up and started working on the design piece. We're using a lot of AI. We're using a lot of the design files that you get with a lot of these chipsets that, that come into the US. We're working with several chip manufacturers [00:32:00] in the US to be able to r- what we're calling re-originate a lot of the design, and we're now able to do a full design of a chipset with two to four people in about three months.
What it has taken historically in China one to two hundred people in two years and $20 million, we're now able to do for sub five. And so that's the kind of work that, that we're doing in manufacturing that's super exciting. We think that cost curve will continue to come down over time. And and it was fun for the engineering team because it's chip turns, right?
So like they all get to dust off their electrical engineering backgrounds and, look at a chip board and a Cadence design file and say, "Whoa, this is cool
Matt: Yeah. Yeah. No it's an incredible story and goes against the narrative that some things just can't be, made here like they were before. I think when we first talked, we talked about the-- I gave the example of toasters, which we had a conversation on this show about, a guest said that we can't make toasters here.
It's cheap or dirty work. And the answer was, if somebody told me I could buy a toaster that was fully made here, I'd buy [00:33:00] the toaster. Now, that's the kind of a microcosm of, where we're talking chips or bigger IoT devices or those types of things overall. But if you're a smaller manufacturer one that doesn't make these types of modules, what are-- what's the example to take from that story about where reshoring opportunities come from?
Is it this, the software layer is the, the capacity unlock that people are looking for?
Andy: I th- to be honest, I think it-- for that company, yes they needed it. Without it they can't cost compete with the rest of the world. You just can't. Now, granted, there's some regulatory capture in the sense that you can't buy from the rest of the world, but at the same time, if a module goes from being $2 to $10 now all of a sudden that's a problem, right?
So you have to cost compete to a certain extent. So they needed the software. I'd say for other manufacturers, we've worked with smaller manufacturers. Software is not necessarily the key. It's more the business operations process. We work with another manufacturer that literally they need... they can't have cell phones on the floor.
It's dangerous. It's an environment that's not conducive to communication via cell phone. Obviously, if [00:34:00] people get distracted, that's really a disaster. But that's why cell phones are banned. So we found a way to bring Android-based walkie-talkies to the floor with kind of AI, so to speak, at the edge, where you could take a picture of something and say, "Quality control, yes or no?
Is this good or bad?" Or you could go over and just talk to it. It would put an order in to the back-end system to deliver materials to your station. So we started doing things like that where that's not necessarily a software platform. That's taking what you normally would have to do, which is have somebody drive around on a golf cart from station one to station two and talk to each other, to all of a sudden now you've got this ability to leverage basically a cell phone, 'cause that's really what you want.
That's the right tool for the job.
You just can't use it inside the environment. Now we've put these, walkie-talkies in the environment that allow people to communicate, and it's all tied into the back-end ERP system. It's tied into the back-end management platform. And yes, there's AI that is interpreting what's being [00:35:00] said.
It's interpreting the images. It's understanding what's happening. But it's just a normal course of business. And and that's a small manufacturer. They're, they produ- they do big, huge things, but they only do three a day. And and they just-- It's a dangerous work environment, so they just needed a solution.
And that's something that we would have never done in the past without AI, because we wouldn't have known how to... we would've needed a full-blown hardware engineer to route the the Android operating system on this random walkie that we get, off Amazon basically. But we've got the software team now that, that's able to do it that leverages AI to be able to build that software that goes on that device.
Matt: Yeah, it's a, it's an incredible opportunity to, again, unlock more capacity. I-- The next phase of this conversation is what we call the Monday morning plan. Our episodes air on a Tuesday, so if it's a really ambitious ops leader, the Wednesday morning plan the day after they hear this episode.
You spent 18 months watching mid-market operators put AI into real workflows. If I'm running a $150 million manufacturing company, I've never hired a software [00:36:00] engineer, how do I know whether to build or buy? Where do I start?
Andy: So start with what you're trying to achieve. What are you trying to accomplish? Like I said, some of the stuff we bring to the table is 2016 engineering. That does not include AI. Really try and understand w- what you're trying to accomplish. If it is I'm running as efficient as I possibly can and I just need an upgrade to my CRM, then, buy a CRM.
Don't buy AI. You don't need AI. But if it's, I'm trying to grow in any way, however you define it, it could be grow profit, it could be grow market share, it could be, take market share at the expense of gross profit. Whatever you're trying to do and map out do I have the team to do that?
The answer's probably yes 'cause you're probably doing most of it today. And then it's if I could give them a magic wand, what would that magic wand do for Andy, for example, to make him that much more effective in his job where now I have four Andys? If I think my team is excellent, how do I get four times the amount of Andy?
You give Andy more hours in the day. How do you do [00:37:00] that? By giving him the tools to be more effective with the hours that he already has. And it starts with, Andy's now one and a half times as effective, now two times as effective, now four times as effective. And start with that rubric, which is if I could give everybody a magic wand and I could make them a little bit more effective every single part of the day, it compounds on itself.
And if you believe that you've got the right team where you give them the magic wand and you give them the a- to get to the outcome that you wanna get to, then, that's how I would start ap- approaching it. Whether you buy software or not, it's like that's the advice that I give entrepreneurs when I was a board member, a, a venture capitalist.
It's build the greatest team you possibly can and then clone them, and that's really what AI allows you to do. It allows you to build the best team. That's on you. Build the best team, and then it's up to us, Jakib, in this case or you as the operator, to either hire a software team or hire Jakib or hire another team like Jakib to be able to make those people a lot more effective.
Because if you've got the-- If you've got team, it's all about team. Just make 'em more effective
Matt: Really good advice and something people can take [00:38:00] away for the next 30 days. We've had a couple episodes on pilot purgatory and people testing, the waters. It's when it's time to execute, it's how do we make the team better and how do we accelerate that? So really great advice there.
Andy, it's been a great conversation. What's next for Jakib? Where can people find you and learn more about your work and what your company's up to next?
Andy: Yeah, so we ha- we are on Twitter, Jakib AI. We are... Our website is Jakib.ai. We didn't get the SI because super intelligence wasn't a thing when we bought our domain name. But yeah, Jakib.ai. I'm on LinkedIn. We've got a LinkedIn page, and just reach out andy@Jakib.ai. Happy to chat. Have you talk to Quentin or one of my engineers and go from there.
Matt: Incredible work, exciting work, big for reindustrialization. Thank you so much for joining us today and letting us learn a little bit more about Jakib.
Andy: This was fun. Thanks, Matt
Matt: To ahead of the curve and to help plan your strategy, please check out our website at veryableops.com and under the resources section titled Trump 2.0, where you can see the framework around upcoming policies from the administration and how they will impact you as a manufacturer. If you're on socials, give us a follow on LinkedIn, X, or Instagram. And if you are enjoying the podcast, please feel free to follow the show on Apple Podcasts, Spotify, or YouTube and leave us a rating, and don't forget to subscribe. Thank you again for joining us [00:42:00] and learning more about how you can make your way.
