How AI Agents Powering the Next Era of Heavy Industry with James Zhan

Being an engineer myself and talking with the engineers, first of all, there is a job shortage when it comes to engineering. If you think about this, since when do you hear people still now when I grow up, I want to be an engineer. Even doctor and lawyer, not as much, right? They all want to be like streamers or Tik Tok stars or whatever that entertainment side of things. If you think about it, industrial engineer is what runs our world, right? Whether it's from a waste, water processing plant to making beer, vinegar, soy sauce to energy extraction. Hi, my name is Demetri Bonichi and I'm a content creator, agency owner, and AI enthusiast. You're listening to the AI Agents podcast brought to you by Jot Form and featuring our very own CEO and founder, Idkin Tank. This is the show where artificial intelligence meets innovation, productivity,

and the tools shaping the future of work. Enjoy the show. Hello and welcome back to another episode of the AI Agents podcast. In this episode, we have James Jen, the founder and CEO of Ranger AI. How you doing today, James? >> I'm doing great. Thank you for having me, Demetri. >> Yeah, thanks for uh coming on the show. Just to kind of kick things off, tell us a little bit about your personal background. How did you find yourself operating kind of at this uh intersection of AI and and heavy industry and obviously a little bit before that? How did you even get into AI? >> Yeah. Um I'm a repeat founder. So this is actually not my first ventureback startup but uh before I went down the tech journey my background is actually industrial engineering specifically in chemical engineering. My uh very first job was

doing um industrial RFPs or tendering for largecale food and beverage company and uh had an idea what the job is but the reality was uh extremely daunting spending all night sifting through thousands of technical drawings, engineering requirements and uh these type of doc. Uh funny story um at one point in my career during that job I was so distraught by the the entire process you know I was 24 25 at the time didn't have time to go on dates cuz every night I was just reading highlighting you know selecting all the the specifications I called my dad and told him I think I'm going to go go back to school and get a degree in computer engineering he's like why you're in a a great field and I told him if this is going to be my life you're not seeing grandkids I can't even

go on this. Um all things aside, um that was actually one of the reason why I went down the computer engineering path and uh hence I'm solving that problem I encountered um here at Ranger. >> Interesting. Okay. Well, very cool. I'm glad that uh we found you going from uh one to the other and this intersection uh makes much more sense for me now. um take us to kind of the early uh days of your, you know, first um ventures that you had prior to what you're doing here. If you can keep it brief, but just like one of what did you do um from a startup standpoint prior to this? And then tell us a little bit about how Ranger kind of came to be. >> Yeah, the two previous venture I've gotten, my very first one was actually very similar to Ranger thesis.

It was using recommendation engine as a way to build out proposals and ERP system for industrial companies. >> Um we're very fortunate to get a set of angels to invest in us. Uh work through the problem and was eventually acquired by a large public traded company. Now I was pretty young at the time and uh um I wasn't able to really go through inventory scale uh and but I learned a ton in the process of how to build a company uh sourcing the right type of founders uh co-founders and advisers and uh that helped me to go through the next couple ventures focusing on Agentic AI for SDRs and prospecting uh and ultimately last year around September I had the opportunity to partner up with 25 Madison out of New York to uh um revisit did my very first set of um industrial thesis and

build out Ranger. >> Very cool. Okay. So, Ranger in a sentence, what does it do? >> Yeah, we help industrial companies um with their tendering management process. Help them win more business, win more bids, and win more RFPs. and tell me a little bit more about what that looks like without the augmentation of AI. >> Yeah, I I'll just explain to my first job. So my very first project I worked on um I was uh in charge of working with my engineering team, procurement team, finance team and uh uh legal team to bid for a massive project uh for the company called AB IMBBEV. >> As you know, they're the owner of Hinekin, Budweiser, all these great beer brand. So, in my mind, I thought I'm going to be playing beer ponds all day on site with my fellow engine. >> Sure. [laughter] But [snorts]

the reality was um I was handed a thick like probably about um uh one and a half foot long of uh thickness in terms of technical documentations and I was told you got to evaluate basically all the engineering drawings count on all the instrumentations and figure out some of the requirements and work with these different counterpart department to gather all the requirements. Um and my goal is to answer a simple question of can we do this which means that have we done a project we have the engineering um specifications or specialties to solve the requirement asked by the vendor and that process took about two months to get done. Thankfully we won the bid and uh um you know went on site to see the construction but that initial process the tendering process was excruciating. Um, you know, not only do I have to review

and work with the legal team on the financial terms, which I have no idea what I was doing, but I have to go through tons of past project to figure out, hey, have we, you know, procured this type of um instrumentations before? Have we done this process before? How did it do before? How do we make sure the margin, the quote, the subcontractor, everything maps out? and next morning the your uh uh professional engineer will ask you, hey James, what's the uh chassis thickness that we need for this part and uh I just got to know or I had to go sift through these thousand page docs and try to figure out where that requirement is. Uh it's not as simple as F because I'm sure you know reference from one document to another drawing to a text spec sheet or somewhere along that uh

entire tender profile. Um the answer lies. >> Yeah. Interesting. Okay. So that's a that's a pretty clear and obvious uh timesaver. I know that um this is kind of where AI I think is is managing to do a lot of work is it's the nonsens sensical but you know you do have to kind of think through it um type of work or stuff you don't have to think through um as well. Uh tell us then a little bit more about how what you guys are doing at Ranger would be described as agentic. >> Of course um as I mentioned earlier this process is a multi-ep departmental collaboration. >> Uh it's not just one person working on their own trying to come up with a proposal or a bid or a tender. um you have multi-layer going from the issuer of the proposal to the EPC

um company in this case um is essentially the the project owner that bids >> uh that flesh out the requirement flesh out the engineering details and then you have the equipment provider or the OEM >> in that process the OEM is responding to the proposal by the EPC's and the EPC is report responding to it by the issuer >> and these issuer are issuing projects in the hundreds of millions to a billion dollars in in in value. So when you go through these type of cross departmental collaboration and cross company collaboration, the agentic side of things will really help you automate and uh reduce not just the timesaving part but improve your accuracy. Um getting that communication layer done and having a multitude of individual digital twins that can help. So I'll give you an example. Let's say um one of my job before is

that I have to go to every single set of engineering drawings and I'll manually count on the design sheet the valves that are on the drawing and then compare that specification to the bill of materials that's provided and make sure there's no mismatch and there's no um uh issues that arise from you know sometimes people make manual mistakes right there's 11 valves on the join but in the bill of interior is only 10 then we have to go back to the uh company I say hey what was the actual answer very manual now you have an AI agent that goes using computer vision to detect all of these type of um engineering specifications create that bomb cross referencing it and provide you with that level of accuracy and risk assessment with a click of a button another example I can give you is let's say

in the tendering process I have to uh actually evaluate uh what are the engineering requirements or the legal requirements that I have for example what's the payment term? Uh what's the actual uh forest manure manure? What do I need to do for um in case you know I screw up how much money I have to pay back uh in that process uh with the insurance policy approved vendor and tons of requirements. How it's done now is someone manually reads through the entire sets of requirement highlighting okay >> and copying it down >> with AI especially with rangers agent agent the AI goes and create that entire sets of um requirement documentations that's customerf facing and generate it you know within a few minutes so when you have these multi- aent collaborating together you essentially will be able to create that uh proposal in a much

shorter time with the help of the agent and the human in the loop just need to review, adjust and uh really um make any necessary judgment in terms of uh uh subject matter expertise or engineering discipline to make sure it's right. Now, tell me a little bit more about how much uh time that's saving um or obviously we're going to talk about more features you guys have, but um how much time do you feel like you save um in a process like that over like a a course of a year for a company if they're like doing a decent amount of work? >> So, for one of our customers, you know, we're seeing I think time saving is one thing, but uh we're saving somewhere between 15 to 30% of turnaround time, which is massive. That's pretty good. see is >> um I would say

a lot of their revenue comes from this is their sales enablement process, right? How do they make money through winning these proposals, these tenders? >> Uh we're helping them especially doing something like 30 plus additional quotes every single uh month. They might not win all of them, but the key is they're competing, they're winning, and it really generate that [snorts] topline revenue for them. And then another way you can think about this is um the ability for you to bid through these massive uh project winning is one thing but the relationship building with these issuer is another thing right I'm sure um you can probably relate if you were to do a a uh home renovation you invite uh three different contractor to do your home renovation the person that never get back to you with a bid you're never going to contact them they're

like, "You didn't even respond to me." The person that responded to you first, even it's a little bit pricier, but if it's done well, they have a higher chance to win. So, when you put that together that if you respond faster with better quality, even if it's pricier, you have a better chance to win. And if you don't respond, you lose that future um proposal opportunity completely. Then I think what Ranger is providing is not just an acceleration of um time to value or bidding on the contract but an overall increase in their topline revenue and a help reduction in their bottom line as well. >> You know that's a very good point. I think there's a lot of problems uh with communication that end up solving a lot of these uh things in so many different aspects of of work. I remember I was

discussing this with a um and it's funny because I don't know if maybe maybe I can you can speak to this. I'm not sure how accurate it is. I think in the industries that are a little bit more I use the term old school. I don't know if that's accurate um but a little bit more like um manual right like less knowledge work uh operationally right like sure there's knowledge work required but I guess the like what they would consider as the output is you know more physical things. There are a lot of people in those industries that don't realize that the knowledge work portion is like where you could probably make a lot of the gains with AI. I had a conversation with a friend of um mine a couple weeks ago. She worked as a I'm forgetting what her role was, but it

was a very small company that essentially did home work, like drywall work, uh, stuff like this for both houses and for >> um, businesses, right? So, it was it was a lot of like it was interesting cuz I I I had to spell it out to her because it didn't come to to in the conversation. It was very funny because she was getting asked by her boss like, "Oh, all this automation and AI stuff's cool. How could I when I have a new client open up like a new folder automatically in Google Drive with some templated stuff? I'm like, this has been a thing for like seven years. That's not really revolutionary. >> Um, I was like, AAPI has been around for a while. But what I what I asked, I was like, so how do you bid on stuff, right? And her answer was

basically like, well, I could I do this. I'm like, well, I'm like, call me crazy, but is this not just square footage plus amount of rooms as like an analysis? Yes. I'm like, you have images of this stuff? Yes. Do you have set amounts of pricing for these? Yes. How long does it take for you to do one of these proposals and put it together? It's like, well, a job opportunity will pop up and then we'll take about like four hours to do it. I'm like, "You do realize you could train an agent to make that happen in like 20 minutes, like at most with you tweaking it and sending it, right?" And she was like, "What?" I'm like, "How many of these do you do a year?" 300. I'm like, "You just I just saved someone a quarter of a job, right? Like that's

a quarter of hours of a work year." So I was like, "Uh, these are the cool things that I think in your industry and industries like it have so much opportunity." So is that kind of what you're noticing? maybe the the industries are a little bit behind in regards to where the technology can help them um outside of like the physical operation of the thing they're doing like the the adjacent things. Maybe I'm Yeah. >> Yeah. No, you're you're on point. Like I've certainly seen a lot of these um proposal coming to the contractors, you know, like whether it's your drywall or your Hback or these type of coding tools, right? That's very helpful. But when it comes to industrial scale, due to the number of players involved, it gets really complex. But at your point um >> the industrial space I wouldn't say they

didn't try but uh um when I first started a similar thesis I mentioned my first startup was in this space um one of the biggest problem I was faced with data readiness and I think I'm still facing that right now some of these legacy companies don't necessarily have all the data that's digitized actually if you were with a a a customer they had a a room literally like a huge room just filled with file cabinets right these are all their old project they have done everything they have done prior to I think 200 and uh 2008 right and you're talking about like hundreds if not thousands of projects that's just not digitized and that's the first thing the second thing is I think AI as a whole on how it uh uh applies in industrial space it's still um being adopted by the enduser themselves.

Some of them like a lot of the professional engineers that I meet are um well in their 40s or 50s. They've been doing the same way they have >> as their mentors and their mentors mentors >> and all of a sudden things that are used to be acceptable that takes about a week or to to to figure out can be done by an AI in seven minutes or five minutes. it it takes a while to really adopt that build the trust with the AI and build a trust with their own skill set on how they can contribute to the value. So >> why do you think that is? I think you know it's like uh it I always make this joke that um if you are used to doing something for so long and that becomes a part of your identity and your job right

like uh the moment I first got a Roomba at home you know I used to vacuum to clean right I was fascinated just by watching the Roomba do its work. Am I fascinated because hey how does Roomba do this or am I more like Is it doing a good enough job like I would? The trust factor, right? So, you can put that in parallel comparison uh in a lot of ways that these engineers with >> that's why I haven't bought one yet. [laughter] >> Um but like if you break it down, there are so many different companies now that's in this Agent AI part for industrials that covers a wide variety. So, we're very top of the funnel, right? We're helping companies to win more deals. We're helping companies to to to earn more revenue top line, bottom line in that sales process. Um, and

I'm also seeing a lot of um uh AI companies now, startups that's doing um process controls. So you can imagine like uh u monitoring the production line. >> Sure. and seeing where are the flow rates, the machine speed, the time, the quality checkpoints and using AI to optimize with sensory datas, demand changes and identify that inefficient inefficiency. That's that's huge. The other one you see a lot is in the quality control part, right? Like let's imagine um you have um with high-speed cameras, sensory arrays and computer visions to essentially at scale detect uh cracks, dimensional variances and all these type of issues during the production cycle. And then you know supply chain inventory management demand forecasting there's another set of uh AI companies that's really focusing on like how do I help your company to contract the raw material provider to get ahead of this

so you don't write into the supply chain issue that we did right after the pandemic or the procurement side um you know a good friend of mine had this company that it would automatically go through all of your um vendor bills that's about to expire >> and automatically start a negotiation ation cycle using agents to be like, "Hey, you know, we've been together for two, three years. Can you lower the price a little bit?" Right? And you know, if you want long-term contract, you have this AI to kind of um pick the best opportunity and stack rank them to do negotiation. You'll for sure claw some money back or do some savings in the process. It's just like uh I'm sure you or me, people like us would uh every few years change our uh mobile provider, internet provider to to save 2030 bucks in

that process. >> Yeah. Sure. Yeah, >> right now imagine agent doing that with hundreds of vendors uh every quarter at scale. So these are all the different type of AI agents that I'm now seeing in the industrial space because the the pipeline the operation is so massive. >> Yeah, I feel like it it has the ability to really impact you at scale uh in a way that maybe some because I think it's it's maybe you know it is complicated more complicated than this. Sorry if I'm limiting it, but I do think um it probably has a lot more like input output transaction number, right? Like there's a higher transaction number than like a lot of service based or knowledge work just doesn't inherently I don't think have as many like transactions. Maybe I'm crazy, but that just sound right or >> I mean they they

do >> like service businesses have transactions, but not Yeah. >> The volume might not be as high as the size and scale of these transactions, right? >> Sure. And you know going back to what you were talking about earlier about this legacy space you know I talked about the data readiness and I talk about the risk and compliance anxiety right like can I trust this to do the right job like if I'm a professional engineer I got to sign up on this AI did a lot of the work uh I don't know if you saw there was a news about this big for consulting company um they they put out a publication they did this work for a enterprise company and then it was making up quotes hallucinating making up quotes didn't exist It was pretty recent, couple a few months ago and uh they

end up having to refund a bunch of their uh bills. Uh and this is why like at Ranger we really focus on controlling hallucination. All the scaffolding we're building on top of our LMS is uh really contained to make sure that we want things for industrial quality which means that you can't have um 85% or 75% accuracy. You got to get as close to 99.9% as possible. And the last part I think the industry is facing right now is change management fatigue. This is something that's often overlooked. um they adopted the whole cloud infrastructure and the whole you know online everything I would say quite [clears throat] recently relative to software companies and then imagine you just went through the whole uh ERP system the whole uh you know ESG rule upgrade and then you went through a tariff shock and now boom AI um

for this latest industry the the agility they they need to have it's not as um trained as software companies. So I would say that has been a very um very challenge that a lot of our industrial agent startup company are faced with. Um and uh this ability to go into the company help them deploy AI agents to help them grow and and get that value um in a a fast um return is super important. So the time to value to customers is pretty indicative of how well they'll adopt you and continue to use you. >> Interesting. Okay. Well, you know, that adoption fatigue is a term I'd actually heard um that you just brought it up. So very very interesting. I um that makes sense though. You know, like they were kind of I mean my I have family that work in other other industries

that are still behind on adoption of other stuff. You know, you mentioned how recently they kind of adopted cloud. Um, you'd be s you'd be surprised to hear a lot of even like the banking industry is like behind on that. Like they still were using manual um I'm getting like like server based cl like uh uh ticker type stuff which is which is wild. Um so we we often are in this position where we're almost in a bubble especially me I'm so big on AI such a power user talk to people like you five times a week. So when I meet people that don't utilize it, uh I see an amazing opportunity um but I also just feel like wow these people don't don't really knows what know what they're missing out on. But then again there are companies who have already adopted something had

struggles there. Yeah that's that's interesting. Okay. Um what would you say is kind of the most interesting or most exciting thing that you're bringing to your product or have brought to your product in the last few months or in the next few months? Um, it's the computer vision stuff >> that that I'm I'm the most excited about. Um, I think it's one thing to be able to work and deal with unstructured text, >> right, or tables, right? >> But another thing to deal with engineering drawings >> okay yeah I was trying to understand what so this is regarding drawings of uh spec. Well, like let me imagine you know going back to that file cabinet of u you know thousand products from the past >> they are handwritten and uh if you are able to say take a photo of these handwritten specs and then

putting to ranger and the AI is able to decipher all that information and uh uh index it properly for you to make it digitize in an instant. That level of power is is incredible. Right? Let's say um I was able to sell this engine I don't know 10 years ago haven't made any modifications now I'm having a massive badge of these engines the person that used to work on this project has quit already I moved on and nobody else really know much about this right you know these type of product have very indepth product catalog has a lot of details right so what do they do right now they open the file cabinet look through the product catalog manually try to figure out okay how how do I quote on this how do I bid on If you're able to just say take a use

your phone, take a photo, put into ranger and instantly create your quote or create your specs or just query for like hey does this specification meet XY and Z. That type of change is super exciting. Um also the ability as I mentioned to detect all the instrumentations in the engineering drawing understanding the flow and the process itself um is very cool. I'm seeing this actually also quite a bit in the construction space in the civil engine space. there's a rise of uh using computer vision to start uh you know creating these type of um bill of materials for let's say if you give it a layout and uh it's able to then tag all the the the steel components and use math to calculate how much it needs whether it's the beam whether it's the um actual um studs and whatnot and create that list

of build materials and that's really cool because right now it's all based on human manually going through this process and verify >> and >> I think that the the that's what our junior uh also focus their time on >> which going back to the other thing I'm seeing in the industrial space is there are a lot less need for bodies like outsource technicians or junior technicians to do type of grunt work >> because I'm foreseeing the next few years more and more of this will be done by AI and the the actual junior's job perhaps will be moving up the chain uh in the sense of start getting more involved with the engineering work understanding the process side of things and uh the need to become a certified professional engineer I think will be even more important um in the near future. >> Interesting. Yeah.

No, I hadn't um I didn't I didn't consider that. That's that's really cool. Uh I I you know it's it's interesting to me you brought up something here where feels like there's constantly a different aspect of AI related to industries that does things people wouldn't expect. Like for me >> images and video and content managed to do wonders recently. like there was a leap of 10% screen uh UIUX comprehension to 70%. In Gemini 2.5 to Gemini 3. That opened up so many opportunities for myself in the content game. What for you do you feel like is one of the recent improvements in AI that's really opened up opportunities for you guys to help out um more uh clients in your industry that you weren't expecting? Um, that's a good question. I wouldn't actually use the technological advancement itself, >> okay, at a a catalyst, but

it certainly played a big part. Um, is the more widely adoption and the improvement in AI readiness that makes a huge difference. So >> because more and more of our users have encountered chat GPT or Gemini or dealing with AI in their personal use cases this adoption into work has been much easier. If you, you know, go back to a couple years ago, you tell AI in industrial, they they can't even like prompting is even like how do I do that? Right? Nowadays you're seeing a lot more um engineers or uh technicians or people in the industrial space that knows how to do a lot of these prompting know how to do these basics because their interaction with um you know the customerf facing LLM and that certainly builds a lot of trust in what AI can do and helps serve as that gateway to

deploy Ranger into these uh legacy customers. You know, the the way to think about this is that if they haven't done something like this before, um you know, if I have never edited a video before, for me to go and start using a tool for content creation, it's very challenging. But if I'm using my iPhone, I have some basic exposure to editing videos, recording on my own. >> Sure. >> Right. Then getting into that at a professional environment, um it's much easier for me. the ramp is much lower. I'm much more willing to adapt to it. So I'm seeing the overall increase in the AI adoption in the everyday use cases in like the civilian use cases you call it actually pushes the motivation to adopt it in the industrial scale uh much faster. H uh you know that kind of brings up something interesting.

Would you say that with AI it's about knowing how to get something done and articulating that in a in a manner so that the agent can properly replicate the understood human process? Um more so than it is about using AI to try to complete processes via it helping you sort out that process. How do you think it kind of plays out for for you guys and the end user? Because that's something I've been seeing more and more. It's like a photographer is going to do killer work with or a videographer is going to do killer work with Sora and the new Nana Banana stuff whereas a or graphic designer whereas like the average person is going to not know how to instruct the AI as much and uh especially with agents and knowledge bases and rules and backgrounds. What are your thoughts on that? So

for us um and I think a lot of my counterparts in the industrial startup world we have this mentality of um limit the amount of openness for our customers. Um the idea is if you can keep the >> agentic interactions to something like comment and output or drag drop and output which is one action from the customer one click you get the output you're looking for. That's ideal. Everything else is a black box. So let's say um you know in Ranger spec uh case like you upload your 10,000 page in uh RFP documents into Ranger. You click on upload the AI already goes through you know cross functional um requirement extraction. It goes through all the embedding of the data. It pulls out all the detail specifications. it runs through the multi- channelannel agents all in the background you know with the manager and what

the user get is the output of okay if I'm on the engineering team here's my focus if I'm on the finance team here are what I need to do um so rather than give an openness of what you want the AI to do the AI tells you hey here's how I get you the output as fast as possible that's most relevant into helping you move fast so that's what we're seeing um and the reason is I am of the belief that whether it's a gentech AI or who knows quantum computer one day right the inner works of how it delivers the result matters aren't uh it doesn't really matter that much to the typical user what it matters is can you deliver value to them fast and can you help them make their job easier and help their companies improve their margins right the ROI

behind it so the more you remove that level friction or the guesswork and the more you get to the detail the better. Let's use another case of Ranger. Let's say you put an industrial contract into Ranger's legal agent. You know, you could have the human prompt something along the lines, can you tell me, you know, all the red lines associated with payment clauses and give me the risk. Why would we want to give the user that level of um openness if we already program a multi-dimensional agent to do that? So they put the contract in there, they click one button and the output is all the red line clause by clause comparison, all the risk analysis, all the action item they have to do and how they can talk to their legal or their customers about it, which is automating their workflow end to end

rather than having them decide on what workflow they use AI for. >> Yeah, that's a that's a good point. You know, I I I have another follow-up question with that cuz I think it's something that I've kind of gotten into, and I'm not saying this because I think work should get worse or anything to that effect. Do you think that there's like a level of um sometimes I do feel like in work we can make really cool stuff? We can make really great stuff. We can make really thorough stuff, but a lot of it is at a baseline, just a set of if then propositions. And honestly, like 90% of what we ended up doing would have been perfectly acceptable and it being done faster might have been better. Do you think that's or better for the context of the situation, not objectively better as

a thing? So, I just want to make that distinction. Do you think that's AI has kind of helped make that something that can be more common in business where it's like I need something I don't need it done perfect. I kind of I need something good. Do you think AI can kind of fill that slot um more so than we think at the moment? >> For sure. Like the 8020 rules, right? Like um I think the AI can get you 80% there. The rest 20 is up to the human to do it. I for sure think so. But um you know having been in the industry for one and a half decade I also want to say like earlier in my career um not just me a lot of my peers my manager stay busy with busy work you know I'm not going to like

make any indictive comments on what I know the percentage of of let's call it busy work is but I certainly have myself done things to stay busy But it's not a good use of my time. I.e. uh according to Dwight >> everybody's done that. Yeah. You're >> more low impact high urgency work or low impact low urgency work. It's just like you just do it just to fill up your time. And I think with AI now a lot of these busy work shouldn't be needed by human and the decision making or aka critical thinking side of the human in the loop I think is more and more important right and that's the part you know I also feel like school doesn't really prepare you for that as much as people like to right um I I think I just saw this research where like 23%

of Harvard MBAs are jobless 6 months or 3 months after graduation. >> How Wait, did you say 23? >> 23. >> 23. Yeah. >> What? >> Um yeah, I'm I I'm like uh maybe I'm wrong on the number. I'm pretty sure it's something stunning like that. You can look it up. Um maybe it's a false um report but uh yeah I'm just looking around 23% of Harvard MBAs have no job after graduating um within 90 days. >> Damn. >> That same number was 10% in 2022. Right. So it doubled in two years. >> Wow. And I'm also seeing like a lot of these um AI companies about like um automating the uh junior job at a hedge fund or financial institutions or compliance reviews or financial reviews or uh analyst research. Right? So a lot of these let's call them entry level job that focus

on the grunt busy work is slowly being uh replaced. And >> well that's been my theory for a while. Um, I remember early on the show is like I kind of came to a realization. It was a bit of an like it was a bit of a two-parter. Like one I did feel like when I first was working in the realm of uh marketing, I was a a marketing associate, right? Um for like just doing basic paid search ads and paid media. [snorts] And I came to the realization because I I moved up and got promoted pretty quickly, but I was like, I don't think it was because necessarily I was that good. I just realized that the job was completely pattern if then pattern recognition and I'm just very good at that, which doesn't mean I was actually good at the skill, right, of

being a marketing manager. But that has planted this thought in my head that a lot of work at an associate level position and knowledge work is just basic if then logic. And back in the day when we had assembly lines and whatnot, now we have the the mechanical capabilities to do the basic if then physical component. >> When [snorts] MCPs started coming out and whatnot, I was like, "Okay, so now we're getting to the realm of APIs are not the limitation endpoint wise." And then we're getting into onscreen interaction stuff which is just like another step closer. So yeah, not even knowing MCPS or oncreen stuff was possible. It's like we have the computational capability here. We just need a a like the end point and onscreen component gets fixed and then like everyone's associate level jobs is cooked in three years. And um I

think I might be on track for that cuz like it's yeah I'm automating a lot of stuff and I have a small company and don't know as much as all these big companies do. So >> yeah and and I think that's also why for me industrials is different. Um again this is just my own thesis being an engineer myself and talking with the engineers. Um first of all there is a job shortage when it comes to engineering >> engineering just since when do you hear people still now when I grow up I want to be an engineer >> even doctor and lawyer not as much right they all want to be like streamers or Tik Tok stars or whatever that that uh um entertainment side of things so but if you think about it industrial engineer is what runs our world right whether it's from

a water processing plant to making beer vinegar or soy sauce to energy extraction. So that's the first thing. Um the second thing is every entry level engineering training can tell you I want to do more engineering work and less paperwork. >> Exactly. >> They all tell you that the paperwork is what uh kills them and that certainly killed me in that career. Right. I want to do the actual engineering work, not so much about being the knowledge extractor for my senior engineer who um you know really relies on me as he would rely on the AI now today. So that's the area I actually find a lot more interest in because um I am certainly seeing more and more junior technicians and engineers having more opportunities to do that so-called the the level up in terms of work and be out of the busy work.

And I do hope um this is going to be a a massive change with all these um gent industrial companies that's coming out because no Gen Z's or Gen alphas now graduating university or getting a degree wants to be replaced by AI. So the only way to do this is coexist and learn how to use AI to enable you. But there's almost no AI adoption for a long time in the industrial space. actually cloud and and these type of new age tech tool which shawn a lot of smart uh talent away from the industry. I'm sure you've know if someone goes into oil and gas out of school like if you're top the class 4.0 GPA from Harvard you're not going to be like I'm going to be an oil and gas engineer, right? Like that's that's just not really the >> you could though

that'd be I mean honestly though like it's not bad. I mean there's [snorts] also a reason these days you know there's uh this article talk about how trades whether it's electricians or um you know contracting for renovations are making record um amount of uh income relative to before it's trades going from you got to grow up and get a white collar job to now like hey trades is great to be in it's just a supply >> yeah parents tunes changed pretty quickly >> yes and I think a lot of that has to do for these uh AI agents in this space. But I do think in the engineering world, it's only going to serve to make things better, faster, and more efficient, which is going to boost the overall um economy like going back to um what we're talking about about onshore and reshore and

all bringing all these opportunity back. Um so that's where I'm seeing and the amount of talent who now can use technology in this legacy industry to innovate. it's actually a huge canvas for more creativity to go um into doing and I think it's a that's why it's very different than some of the other fields um you know that's very very easily automated. Yeah, fair. No, that's uh that's a good you know and and did you coming from that previous engineering background uh or not sorry not previous pre Yeah. Wait, did you coming from the old background you had getting into what you're doing you were doing now more on the tech side. Did you kind of notice some of these things and patterns uh prior to getting back into the tech world and maybe I don't want to, you know, try to make you seem

prophetic or anything to that effect, but like did you see any of these trends kind of early on and that's where you're like, you know what, I'm going to get into this because practically speaking, it's it's going to be beneficial to me cuz I would love to honestly it's crazy. I think we're the first generation of wild that would love to get back into the more brickandmortar uh like physical type stuff again because I'm like I wish I could buy up some boomers business that's like doing selling some widget or whatever. Yeah. >> And it would be >> it'd be pretty chill in comparison when I get the admin work down. >> Sorry. Anyways, I have yo this is a great question. I have something very cool to share this. So my wife, she is also a business owner. >> Okay. He owns a like

Olympic level hip-hop dance studio for kids. >> No shoot. Okay. >> It's as brick and mortar as it gets, right? You got like a physical level dance studio. You have like hundreds of families and their kids are aged between like five to 15. And [clears throat] what I realized just watching her grow her business, you know, first of all, she uses a ton of AI. It's crazy. She automates everything, you know, from billing to um customer success to like uh event management to content creation. But the one thing that's very different in the digital world versus the brick and mortar is the community. M >> she has tons of these interactions with end users in a face tof face environment and being able to cultivate that level of like almost that tribal community like method that it's very hard for us to do even if

I you know zoom or or or chat with you online multiple times that uh inerson interaction continuously in a um brickandmortar surrounding you know it's hard to replace right And there's also the reason why people feel more detached working from home versus in the office. >> Um, so I'm I'm with you on that. I think the brick and mortar business is definitely making comeback and I think with the adoption of AI in this space, we'll able to see much better margin. Um, and you will make a business much more viable. Um, you know, to your point, you buy out a boomer business, you then apply a set of automation to it, you'll instantly get that level of return. And I think that's what a lot of uh private equities are trying to do right now. >> Yeah. Yeah, I mean I've heard this thing this

twostep where people who are like decently successful, maybe sold out of some other company in their 30s, um are now like buying businesses that boomers are retiring uh from and then operationally improving it and then selling it to private equity firms is kind of like this interesting two-step that is occurring um with successful uh you know 30somes. >> Yeah, I have a friend who recently bought a company that manufactures um fireplaces >> yeah, that sounds like an exact perfect thing. I'd be like, "Oh, there's so many things I could do to to make this more efficient that like they wouldn't have even thought of." >> Yeah, they have about um 10 different product that they always manufacture. So they bought the company um took out a a business loan and then the revenue for is about 1.2 a year and the margin was pretty bad.

the margin was like >> no 15%. 15% is >> and uh and then they dug into the numbers. Tons of it is just due to paperwork inefficiency like mis collecting bills, miss billing, like tons of these type of stuff. >> Follow-ups. >> Yeah, the follow-ups and obviously the material costs, right? They because there's no like good forecasting, they would uh um >> just buy >> Yeah, just buy when it's necessary. It's like, you know, you don't have any >> you can buy in bulk, I'm guessing, and save money. You could Yeah. >> Correct. So yeah, they came in, they literally used um Gemini ChatGpt and then uh and they built out their own website with like auto configure like really quickly using you know tools like um Stripe and Shopify. >> Yeah. >> And they pushed their margin from 15 to about 32% in like

six months. >> Yeah. That's crazy, >> right? And then they put on social media. They start like talking about look at this cool fireplace. >> Yeah. Then they got more revenue. Yeah. >> Right. It's just it's just like I mean admit it it's easy the day they put their blood, sweat and tear into it right all day every day. >> It's a lot of work but it's it's work that the other c the the other owner was never going to get to because that wasn't like where their head was at at all. Yeah. >> Correct. So they they they now boosted their revenue by like 50% and increased margin to 34%. And that's huge, right? Like >> incredible. Yeah. >> You know hopefully they will get an exit in a few years but that's the power of automation. >> Absolutely. Well, you know, just kind

of as we get closer to the end of this episode, what would you say, speaking of cool stuff, is your favorite personal AI tool right now that you're using? >> My personal AI tool, Rooflow. So, yeah. R O B O. >> I've not gotten the same answer more than once. That's funny. Okay. Rooflow. Everything you need to build and deploy computer vision apps. Okay, got it. >> Correct. It's It's so cool. Like, you know, they've been around for a while, right? But just the idea of anyone with enough data and data labeling skill can build their own computer vision model just opens up so many cool use cases. You know, there are some stupid ones you can think of like civilian use cases. You can make a a fun app or a fun uh um meme out of it, right? It's just that yeah, I

I play around like for work of course, but personally like I play around with it just on random stuff, you know? for example like uh um and there's there's tons of uh things if you have enough image data you can just label it uh for do certain things you know >> which way have you ever seen the show Silicon Valley [laughter] >> of course [snorts] one of my favorite >> hot dog not hot dog >> yeah >> it just immediately came to mind Jinyang didn't have to do all that work >> he could have he could have made seafood oh that's so >> yeah you train you train AI to recognize all kinds of complicated stuff, right? And um you know, like uh uh it's funny because I was showing my wife some of the random stuff that I was using work and she's like,

>> "Why do you spend your time doing this type of stuff?" And I'm like >> she's like, "What are you doing, dude?" >> Yeah. It's it's it's like, "Okay, there's the work side, which I really enjoy, but I feel like this is the fun part of things, the experimentation part." Like you can't >> Yeah, you got to have fun with it at some point. Like you can't I I think that I was telling on another interview earlier I was saying this is to some extent a video game. >> Yes. >> Like all this AI stuff. >> Yes. You know uh I saw this one um college kid that built something super hilarious. You can put in your textbook and then you can pick the game mode. It'll extract all the core content in the textbook >> and make it into a game. Hm. Whether it's

like, you know, like uh you know, you got to solve this equation by shooting the right answer in the sky or like like an RPG. You got to like figure out like talk to the >> Yeah, that's so smart, >> right? And it's um it's all AI generated. Like how well it works, you know, hard to say, but it it certainly transformed like when I was a kid, I can remember when my teacher made education fun for me. I'm much more inclined to learn, right? And uh funny enough, a lot of my Russians I learned uh playing Red Alert when I was a kid, right? So, so the same concept can be applied to like um these days, you know. I think uh that's where the the fun part when uh deployed properly is going to to um you know, really change a lot of

the fun foundations of of our society. >> Yeah. Um >> I'm curious what other answers you got uh in terms of tool. I'm just >> uh I have been giving one consistently that's crisp AI. It's a um one I actually saw on the show where basically it's a 0.5ish second delay but it transcribes calls and uh gets all the background noise out. So say for example a lot of busy business owners or like founders whatever like on the go in their car taking calls and I'm like I think if all of them just had that it would be so much easier because they could I don't know uh we can get into whether I don't know if the API is well built out but if it was well built out all these these nerds could one make it sound better so people are like oh

he's in a professional environment his camera's just off or and sorry and or automated post call workflow stuff. I always think transcription is ideal and it's one of the few ones that manages to do it I think even if you're on mobile because a lot of those tools require like a browser extension or like a in meeting bot. So it doesn't have an inme meeting bot but it's managed to to be multiplatform. So that's one of my favorites. I love text core text. It's just my favorite like LLM uh rapper. Um and I've heard a lot of interesting ones over the the last couple um weeks. I started to ask this question about a month ago and um there just a bunch of there's uh there's that one voice one I'm forgetting the name of it where it's a whisper uh whisper not whisp not

whisper sorry this >> it's not whisper yeah no it is whisper it's the one where you basically can install it on your computer I think and or is it whisper flow yeah wispr flow it's the one where you can basically install it and it's a transcription so you can just kind of mouth vomit because honestly man like with agents. Now, if you know how to put agents together, you [snorts] can just be making lunch and explain as much as is needed and make a little mini agent for whatever task, put it in an automation, in an hour or in 30 minutes, you can have like a solved workflow about something. It's very It's incredible. If you just transcribe it, then parse it with Yeah. >> Yeah. I saw this one engineer um it's just a funny um flex. He basically said like, "How do I

be a 30x engineer without touching the keyboard?" >> What a I love this man. >> He built like 20 different agents and it's all voice controlled, right? To basically build the requirements, call the agent, evaluate, bug, um, test out the feature. And all he had to do is like click on the the he's like, I don't want I just want to click on the the, you know, the commit, you know, the PR uh, review process and get it into production. Um, but it's incredible, you get to that scale and it's totally doable that you just use your voice. It's like um the first or the second Iron Man when he's working on his um suit and the whole time he just talking to this AI on collaborating on how to build a suit, right? It seems wild how crazy it was back then, you know,

when Iron Man came out with >> But we're kind of there in a lot of ways. >> We're kind of there. Yeah, >> you can crack the car. you can have a full-on conversation with your AI and then just uh build out your your your week, your workflow, your OKRs, your planner, whatever you're looking for, right? Just through a conversation. Yeah, systems thinking plus agents I think is one of the single and um audio is one of the if if someone wants to get ahead in this world is like you need to understand like basic bas basic system and automation architecture learn how AI agents can ingest data um make rules that kind of stuff and then find a good workflow to take those first two things and just mouth vomit your knowledge and then you can automate many parts of your job and your

honestly your life too This is the fun video game part and I'll then I'll uh trying to close things out. We actually had a such a fun combo. It went longer than I expected. Uh the my girlfriend was talking to me about how she had a get together planned with or wanted to plan a get together with her and two other people, but like she goes back and forth between the Chicago suburbs and the city proper. >> And someone was in the west suburbs, someone was in the north suburbs, someone was in Chicago. and she's like, "I wish there was a way for me to triangulate the best options for us at a reasonably close amount to multiple people that are not just two people." And I was like I like went in lovable and was like I was like Google Maps Matrix API just

like typed set up an API key in Google. Next thing I know I'm like here you go. you know, I'm like, the algorithm the algorithm is going to be wonkyish, but then like I I played around with it too long, but in four hours it was like genuinely good. Like I don't think it could make any money really, but like it was for personal use. It's like, you know, it's kind of cute that you're able to play around and >> it teaches you how to talk to AI, I think, even if you have stupid side projects that go nowhere. >> For sure. I mean like to your back the personal assistant agent like the ability to craft this is just um it's the power you can only dreamed of just five 10 years ago right like I um I had my own like gym agent

workout agent you know that it's like I basically follow these yeah influencers on YouTube right I just uh get some of their transcript and then put that in there and then say hey this is the work uh workout plan that I want I'm going to update it all this time and create the rules and then if I slip, you know, you got to yell at me like uh you know certain influencers online like and [snorts] it's it's hilarious like um >> just have it talk like Gary V did. >> Like seriously, I just want some dude yelling at me about like uh grinding in blueberries and I'll probably like never stop working. >> It's like a military drill sergeant that's constantly bringing down my neck, you know? >> Um >> yes, that's who I was thinking. All right. So, just to close things out, where

can everyone go to find you? Um, and, uh, check out everything that cool that you're doing over there at Ranger. >> Yeah, you can go to ranger rfx.com. Um, or just search up Ranger AI or Ranger RFX um, on Google or LinkedIn. Type in Ranger AI. Um, we're pretty easy to find. Connect with me on LinkedIn. Happy to um, show you what we're doing. Um, but more importantly, Demetri, thank you so much for having me. This has been a lot of fun. >> Yeah, absolutely. I've had a a blast this entire show. We appreciate you for coming on. And if you guys like if you guys enjoyed it as well, make sure to go to the uh Ranger I got to say it correctly. Ranger RFX.com. That's ranger with uh one r and then another rfx.com. And also hit the like button on this video

if you're watching on YouTube. Leave a review on Apple Podcast and Spotify to get this out to as many people as possible. Thank you so much for listening or watching and we'll see you in the next one. Peace.