Many are bleeding money — and most construction leadership teams don’t see it until cash gets tight or the bank starts asking questions

(00:00) Good morning, everybody. I guess it could be morning, could be afternoon, could be just about anything here at the Contact Exec Podcast. contact exec podcast but welcome everyone who’s listening in and i want to welcome my new guest uh rishi shrivastava i do have my texas accent has a little trouble with that but i try to work my way around it rishi thank you for beiing with me today how are you doing doing well thank you robert friend yeah very good well you and I had a conversation.
(00:45) It had been a couple weeks ago now just about your company beiing human and talking about the redundancies of some of the work. I just want to – how did you – and I always ask this question. How did you get into this crazy business when we start talking about, you know, construction and all the paperwork around it? How did you get into this? Yeah, I was working at Bank of America and I was living in Charlotte.
(01:18) I met my good friend, Luke. He’s a construction CFO. good friend luke he’s a construction cfo and i was working as an ai engineer at bank of america as well during that time i myself was doing a lot of tedious work looking at one report in pdf and tapping stuff into another place uh kind of conversion from the some structured to structured data and Luke was a very forward Luke is a very forward thinking CFO and he and I had a great chat and he was like in construction there is so much labor wastage on these documents information from for example from
(02:02) an invoice extracting it out and putting it into an ERP, such a tedious task for an AP clerk. And some of these companies, they have people, maybe a couple of AP people who are just typing this information in, AP invoice, delivery ticket, credit card receipt, all those documents.
(02:29) So the name of the company, going back to your original question, came about the idea that humans are not supposed to be doing these tedious, repetitive tasks that an AI can do. And yeah, that’s how it came together. Yeah, it’s probably a little bit below folks, you know, kind of pay grade, I guess, or intelligence grade to be having to, and there’s nothing more frustrating than having to do redundant work, like, you know, extra entry.
(02:58) And it also introduces the element of error. As much as smart as we are, right? We are humans and we do make mistakes and we do tend to lose focus from time to time. I know I am very, very much susceptible to that thing where I lose focus for a little bit. Maybe, you know, screw a few things up. And I get it.
(03:26) Nobody’s, nobody’s adverse from mistakes, but I think getting to where we’re at with construction and, you know, what we’re doing, it’s very important to realize that those mistakes cost dollars and cents, right? And that costs time, I guess. And if you’re going to relate both of them, time is money, whatever you’re going to do. But they tend to compound on one another.
(03:53) One mistake maybe leads to two, maybe leads to more time, maybe leads to that. And so you just decided one day, okay, well, I’m going to jump out of Bank of America and I’m going to help my friend out. And we’re going to talk about this construction industry. Bank of America and I’m going to help my friend out and we’re going to we’re going to talk about this construction industry because I you know I have called people crazy before about getting into this industry um because it is so complex but so is that how it happened you said okay yeah I
(04:14) think we can do this yeah so looks definitely had faced this problem himself. And this industry, like you were saying, it’s a very complex industry. Furthermore, people are not willing to change easily too. Something’s working, and even if it’s working really badly, the risk of change is just sometimes too much for people in this industry.
(04:44) Yeah, isn’t that funny? It’s almost like the devil you know, right? It may not be working great, but it’s kind of working. And I’m afraid that something else that I might try may not work at all. It seems to be a bit of a superstition almost at some levels. And maybe it’s because it’s not the background.
(05:07) It’s not really a core competency that is within the industry. But a guy like you, coming from that Silicon Valley mindset, for you, it’s like, oh, no, no, this should be much easier. Yeah. I mean, we’ve been doing this for about four years now. And first couple of years, people were not even listening to us.
(05:35) It was just so hard to get people to pay attention. Welcome to construction. Yeah. So tell me about your Silicon Valley days a little bit. You were kind of in the early, you know, kind of applied AI onset. What did you see, you know, kind of good, bad, and ugly as you kind of made your way through that? In Bank of America, we were already implementing things that became very popular around 2022, 2023, even in 2019.
(06:08) Our consumer AI domain, we had a chatbot called Erica. I worked on its intent detection and entity detection portion of it. So when you’re talking to our chatbot, for example, you say, show me transactions from yesterday. The chat bot needs to figure out what you want to do. It also needs to figure out what are the important things, entities in that kind of question.
(06:37) So developing models for those and we were developing things that were really tuned for our data set and and not to the general you know chat dpt or current generative models they’re very general but we were working on something quite specific to make sure that it worked just for the bank and during that process from 2019 to 2022, I was like, why is this stuff just sitting here at this corner of the bank? And why is it not everywhere else? You know, and that’s the kind of mindset that helped me make the jump.
(07:15) I got you. So you bring up an interesting point. The information that you were dealing with at Bank of America was very pointed, right? It’s very specific to Bank of America. Just had the information that we posted on. I posted on last week. AECOM just bought Consigli for like $630 billion or some crazy amount, a million.
(07:43) I can’t remember what it was. for like $630 billion or some crazy amount, a million. I can’t remember what it was, but basically bought themselves their own AI company to put under there. And then you also read the stuff about Turner who decided to contract for two years with ChatGPT, which again, I’ve commented on it about it beiing, is it a strategic move? Obviously, it is some sort of strategy to move into AI.
(08:11) There’s two different schools of thought on how to get there. My concern with either one of those is kind of current processes, how people are doing things. Kind of like you said, folks wouldn’t even talk to us for the first couple of years about this kind of stuff, even though it may have been causing them pain.
(08:31) They just couldn’t get around it. And I concern myself with this because data governance is probably one of the biggest things that we have a problem with. People call things different things, And people have different processes, even within the same company. You have different spreadsheets. You have this, that, and the other.
(08:50) Well, I do it this way because I had to solve it this way, and I don’t want to move away from it. So what’s your thought on that? I mean, the chat GPT route, as you said, it’s very general. It’s very widespread. I mean, are we teaching it things from you know somebody’s blog post um i mean i know we give a lot of great information here on the contact exec but i don’t necessarily know that you know they should be quoting me on chat gpt at any point in time uh in the turner infrastructure I mean, I don’t know that that would be a very wise thing to do.
(09:26) So again, isn’t the more detailed the information, the better data governance we have better for an AI functionality? Oh, yeah, especially for huge organizations like Turner. You don’t want like finance P&L showing up in a salesperson’s inbox, right? The data governance is big in big companies. And who owns the data? How can it be moved from one place to another? What sort of querying you can do on that? another, what sort of querying you can do on that, who can actually listen to the reply from the bot on that particular set of data.
(10:12) All that’s super important. And going back to that, these acquisitions, these construction companies, the first one that you’re talking about, and suddenly getting acquired by a construction company, I’m not sure if it’s the right move you know the construction companies their core expertise is never technology I mean if you’re trying to be a technology company uh you know there’s plain vulnerability maybe they want to be a technology company you know and then you because if you’re acquiring a company just for yourself it just
(10:46) if you’re acquiring a company just for yourself it just that’s not the best idea there’s so many so many other ways to use the money you know but maybe you’re acquiring it and then you’re trying to become a software company yourself and you think that you have that distribution right but the chagi period and uh and the trainer partnership it kind of makes sense to me it’s not a kind of uh telling is a huge empire we’re talking about right i mean not talking about a small company and now in u.s the smb market is huge too right i mean under 500 million dollar revenue there’s a lot
(11:17) of companies and those companies uh i mean they cannot directly get served by open AI. And one more thing we have to remember is I want to time this AI workflows. They need to be very tightly integrated with your current processes. You know, they cannot just come and sit in a silo. And how are you going to make sure that the workflow is flowing post adoption? Yeah, I was concerned about that too with the chat gpt thing um you know each body somebody using their own gpt that they’ve created that becomes siloed
(11:56) information again unless they didn’t share you know becomes a system gpt which it’s shared amongst people and is everybody going to use it um you know the adoption um I always say there’s two different things there’s adoption and there’s adapting right they’re not they may come from the same you know word but it’s not the same thing um but I kind of want to get into more about you know beiing human um and and what you guys are doing with the you know the AP side of things we can talk all day about the more about you know beiing human um and and what you guys are doing with the you know
(12:26) the ap side of things we can talk all day about the way that you know some folks are trying to adopt i’ll say ai uh tools and whether it’s the way turner wants to do it by using a chat gpt or whether it’s the way that maybe a ecom wants to become a software company however but a product like yours it’s it’s a dedicated tool right um and and it’s dedicated to helping specific uh pain point and i want to talk about that pain point which is kind of like ap and the redundancies and and why is it so expensive
(12:59) what what is going on within those systems that you’ve discovered that makes it so expensive and how can AI help? Yeah. Think about an accounts payable clerk and maybe they’re receiving 500 invoices a month. And they got to receive their invoice and then type it in the accounting system, get the approval from project managers.
(13:28) That process itself is very time-consuming. Getting one invoice into the system can easily be five minutes, you know. So we’re matching it up with a purchase order, right? What if it doesn’t have a purchase order number on it, right? Yeah, that’s exactly right I mean there’s just so many uh scenarios you know pure number missing like you were saying maybe it needs to be also lined up with a drill ticket you know we also tackle that right so how how do you uh make sure that there was delivery against that AP invoice that you’re ready to pay for you know one
(14:05) aspect that a lot of times gets missed is the approvals from the project managers the current systems they’re not web-based pretty much you know we’re talking about old ERPs in in like built in 20-30 years ago and it’s their UX is quite old and a lot of times people don’t get a entry and then when things get into a bad part of the workflow for example the project manager rejects and annoys maybe it was not for his job.
(14:45) And now suddenly that part of the workflow is not very well handled in the old ERPs. And the information gets lost deep inside these complicated interfaces. That’s where another opportunity was there for us to make sure that information flow amongst various teams, field, project management, and accounting was happening smoothly in AP.
(15:12) You know, it was not just about getting the data typed in and matched properly. It was also seamless collaboration amongst various parties here. Yeah, I would think, I mean, I remember, again, back in my good old days of running projects and getting invoices that were incorrect. So the invoice number was the same.
(15:36) It’s all there. It’s all in the system, but it’s incorrect. So then you’ve got to change it. but it’s incorrect. So then you’ve got to change it. And if not everybody is in the same loop, right, they’re going to be seeing maybe in the system it gets duplicated somehow. Like, oh, there was this invoice, but then they sent me this invoice and this invoice. Which one is the real invoice? Because there were changes made to it.
(16:03) And so I remember sitting, specifically sitting with AP clerk and sitting there with them going, okay, this one is the correct one. We had to change here, here and here. And that, you know, and those systems, again, like you said, we didn’t have an integrated system where everybody could be on the same page or be notified of changes made to specific invoices, purchase orders, documents, anything like that.
(16:26) And one small change can really throw a wrench into the whole works because then it may look like you only have, you know, it may look like you have three invoices, but you actually have one. And it gets very, very confusing which one gets paid. And it looks like you have approved them. And if you don’t all have all of that together in a work order, especially an organized, or like a workflow, especially an organized workflow, you end up in trouble very, very quickly.
(16:57) Very well said. One more thing we’ve observed is the cash flow visibility. A lot of times, you know, these AP invoices, they’re sitting outside of the system. People are having all these folders on their Windows machine and people are, the PMs are approving in these folders and the controller or CFO has no real clue about the liability they’re holding for 30 plus days you know until people get to code it you know and get it in the system right that’s another thing it’s it’s the notification of what
(17:31) is changing within the system as it happens it’s to your point it’s kind of real time because it’s important for a cfo or controller to know oh oh, hey, by the way, this is cash going out and this is when it’s going out. Oh, no. Did that change? Oh, OK. Well, how much cash was it that’s going out? I’m not just talking from the AP side.
(17:54) I’m not even talking about from the accounts receivable side, because I think it’s probably much of the same thing there where, oh, yeah, by the way, they were supposed to pay us, you know, on this date. And now we’re not going to see it until this date, according to such and such. Right.
(18:16) And again, one mistake, one mishap throws a wrench into the entire process, because not only do you have to focus on that problem. Right. But you’ve got 50 others that are going to come in this week that may have the same situation. So tell me a bit more about how beiing human kind of does some of the heavy lifting. One main thing it does is the progress on AP process, it’s available to all the interested parties at real time.
(18:45) As the changes are happening, the AP is getting the information into the system. PM is approving it or rejecting it. All that flow is happening seamlessly. And audit trail, the life cycle of this transaction, right when the first time the AP invoice came into your inbox, there’s a lot of things happening.
(19:14) People are quoting it. There’s approval rejections happening. Maybe someone attaching a delivery ticket. Maybe you’re assigning some approvers or there was some issues with the vendor’s invoice ID. All that stuff is getting logged and people can see that in the audit trail. Maybe this invoice took a total one week to get processed to our system.
(19:40) But meanwhile, everything that was happening with this invoice, all the conversation is centralized. And one more thing we do is uh q a on the uh documents so and let’s say an ip noise stands into the inbox of the accounts payable person they have no clue sometimes whose invoice is it right they don’t even have a good starting point so they can ask questions around that particular invoice instead of starting to route it randomly to various people and try to figure out uh who’s invoice that is the kind of the chain of command almost on who’s in charge of this this budget this this amount
(20:19) of money those types of things i got to thinking about that um i think again notification of what’s happening with certain files um is a big deal but then it’s that redundant factor because are are you guys utilizing your system to basically create uh does it does a po start in your system or is it something that starts the PO starts in the ERP where it kind of should be you know that’s main financial and you guys are just monitoring that system and making sure that once it’s in there then now it can transfer easily into its next stage and match
(20:57) up with the invoice is that how that works? So our customers sometimes they have POs in the ERP sometimes they don’t some of our customers don’t even have POs in the ERP, sometimes they don’t. Some of our customers don’t even have POs. They just deal with APNOs. They’re just maybe not there in their maturation lifecycle.
(21:15) You know, maybe 10 to 15 billion in revenue. They just have not gone through the PO process. But some others have POs too. When you have like around 50 million or 60 plus revenue, that’s when POs and all these controls become very important. So for that type of scenario, the customers can create the POs in the ERP.
(21:36) Sometimes there is this purchasing system that they have, where that is the one that takes care of all the purchasing and pushes the information into the ERP. But if the customer doesn’t have that kind of purchasing system or they’re just creating in the ERP, that’s okay too. One more thing we do for some of our customers is actually create the PO too.
(22:02) Let’s say the vendor is giving you a quote, you just bought out of this job, uh, you can start with this vendor quote, uh, or field order from a field person. And you can let the AI read that and, uh, code it and push it to the ERP too. I gotcha. So the AI layer basically acts as that communication layer that’s missing, um, within the current system or can even perform on its own.
(22:31) So for us, the AI layer is mostly a data transformation layer. So when you look at the AP invoice PDF or a relative ticket PDF, a human can look at all this information how do we translate into a structured format uh a json that can be read by an erp that’s the main year earlier we have we have a small amount of other ai layers for example um we have a chatbot which can help you navigate our question if you have any concerns questions about this our system you know that’s another ai system mainly if you see think
(23:10) about our system we are just converting unstructured to structured data and that conversion it has to be template agnostic too because different vendors have different type of ap invoices they don’t all look the same. You know, if you have to start setting up the vendor templates, that’s going to be very miss-messy, and our AI does that very well.
(23:36) And one more piece to that I’m going to add to it is validation layer. How are you going to validate AI’s work very well? And we’ve spent a lot of time developing a statistical user interface whereby you can easily see where the information is coming from in this document through our user experience user interface right is that what you call responsible ai um when you’re talking to folks about is is that? Yeah, that’s the right way to think about it.
(24:06) Yeah, I mean, how are you going to validate this AI’s work? That’s always a question anyone who’s considering AI in their current processes need to think about. Yeah, well, hey, that’s a great point. I’ve talked about it several times on the podcast here, which is you can’t treat AI in its current state as plug and play, right? You have to treat it as kind of a new admin or a new project engineer, you definitely have to check its work because it’s not perfect.
(24:46) And I don’t know that it ever will be perfect. I had a conversation the other day to thinking about large language models. Are large language models kind of capped out? Because there’s a lot of talk about that in AI circles, circles how llms may not necessarily be our best path forward when we talk about the overall overarching topic of of artificial intelligence right maybe there’s a better way maybe a better engine i wonder what you thought about that you know if you’re thinking about artificial general intelligence agiGI, I don’t think
(25:26) algorithms are the path. These things are just a statistical pattern in some ways. I mean, they can talk very well, but they don’t have very good understanding of the outside world. You know, physical intelligence, they lack big time. Also you know, they’re a lot of times just responding to you, trying to cooperate with you.
(25:49) They’re not trying to get really at odds with you. Whatever you say, they’re going to reverberate back to you the same sort of sense. Yeah, they’re too nice. Yeah. I love the word they use they’re trying to cooperate with you there and i think that’s important to know and it’s important to note that as i again if you get on a on some of these especially when we’re talking about like llm chat dpt that’s what i was worried about a little bit with you know like turner’s idea of you know we’re going to give everybody in our company or whoever however however many folks are giving it to, I don’t know. But for the next two years, you know,
(26:28) we’re really going to let them develop their own, you know, GPTs and, you know, see what, see what that develops. I was reading, I don’t know where, if the statistics are true or if the numbers are true, but they’ve already developed some 400, you know, different, I’ll say agentic type.
(26:47) I don’t know if there’s true agentic layers, but agentic type GPTs amongst their current staff. And again, I need to do some better research into that. But then I start to worry about those things because as to your point, they may not necessarily be the, they’re really smart, but they’re maybe too cooperative, right? They want to – I don’t know.
(27:10) They want the attaboy kind of thing. It’s almost, you know, oh, I don’t want to upset you and I don’t want to hurt your feelings, but yeah, let’s try that instead, you know, even though it may. Again, reasoning, I guess, is one thing that we’re missing in all of this thing. Statistically, we can say that a statistic is not a reason for something, right? A statistic says, well, over the last 10 years, we’ve seen this result 70% of the time right mm-hmm so that may not directly apply to your query whatever that is but I’m gonna tell you well this
(27:53) worked for this so let’s but again if we don’t have a one-size-fits-all type of world yeah yeah and I think that’s gonna be the biggest stumbling block for artificial intelligence, especially in the construction industry, because there’s nothing about the construction industry is that one size fits all.
(28:12) We have several layers of, you know, parallel processes, right? You know, everybody, yeah, we’ve got some framing involved. We’re going to hang some steel. We’re going to pour some concrete. But there’s so many different variables within that. I don’t care whether it’s weather, location, altitude, those types of things all affect how well and how productive we are in a construction business.
(28:45) That’s just the construction portion. I’m not even talking about the stuff that we’re talking about here, about getting the bills paid, right? How many hands are touching certain products and materials before it ever even gets to the job site? How much paperwork we’ve got to go through to get there? Let’s not even get into the pre-construction phase, right, of how well we’re estimating this project how well we’re doing on these so there’s so many different real world variables that i don’t know that artificial intelligence in itself is certainly at this level
(29:19) able to handle such things i don’t know if it’s a good idea to use a predictability module on something like this yeah i, you have some valid concerns. One big thing is that you’ve got to treat it as an assistant. It’s not your boss. It’s verdict is never final. Yeah. I think it comes down to this.
(29:44) It leads to a whole new level of, I don’t want to say it’s supposed to relieve stress like when they talk about ai when we talk about utilizing it i don’t care whether where you’re using it whether you’re using it at bank of america or you’re using it at turner construction or you’re using it anywhere i don’t care you be a mom-and-pop shop down the street there’s a different sort of stress now it’s supposed to aid you in your process of thinking in some ways it does sometimes i think it might maybe we can’t let go of our critical thinking right so now we’ve got another we’ve got a different stress layer we don’t have the stress
(30:17) layer of going am i thinking about this correctly what now we have is um i’m looking at it this way my ai assistant over here is looking at it this other way is this the correct path right and leadership is expecting you know if you work for somebody and you’re utilizing ai leadership is expecting that to be investigated as well they want to know, well, how did you arrive at this answer? So you mentioned, you know, leaders drowning as AI scales faster than governance does.
(30:57) And I think I want you to kind of address that because I think that’s a big deal. Yeah, and it cannot be an afterthought governance, right? governance right i mean we kind of have to know that uh at the very outset this thing where the data governance it’s going to be big um has to be kept in mind right at the outset you know like when you talk about a big project like let’s say you and i are building a big building right we cannot be thinking about the very last step of uh building that building when we are there we have to think about it like right at the very
(31:33) inception of it when we maybe even when you are estimating it right you know so data governance is just like that you know it has to be taught right at the very inception of the project. And it has to be used throughout by everyone. Data governance is just that. And I know it’s a big word in a lot of people’s – I mean I know when I first heard it, I probably wanted to fall asleep.
(31:56) Because, again, it’s not something that you – it’s not an exciting process. It’s not an exciting process, but I think to you in this term, I almost consider it the rules of language. How we speak to our AI agent, our artificial intelligence, how we speak to it has to be consistent and it has to be structured, right? Because it’s not any different than you and I speaking different languages.
(32:34) I wouldn’t be able to understand what you’re saying because I don’t know your language. I don’t know those rules, right? Sure. Just the same thing as you understand, you know, you’ve learned English and you probably have a second language um right we have different rules for english than you do for uh your native language those types of things and i see it as the same problem with ai i mean early on people started talking about you know early on they’re talking about the system the prompting you know prompt engineering is the new thing i think that’s a it’s a bold word to say that you’re engineering some sort of prompt
(33:09) but most people don’t understand what that means but that’s all part of this governance package and then when you say governance package you know a lot of people just tune out but is it is tune out but is it is um the most important thing i think as we enter into this what we say age of ai the most important thing that we can talk about is let’s communicate properly with it so that we’re not creating confusion from the beginning of this yeah yeah so does this create new roles within companies do you think you know it definitely You know, we live in a world where people are not fully
(33:53) ready to adopt the technology. And there has to be at least a few people in an organization who are thinking about data governance. Yeah. Yeah. Well, and if they’re thinking about it, they have to be able to teach it. Because again, you’ve got, you know, just like somebody who is a translator, it has to be able to translate.
(34:20) You have to be able to take somebody’s information and say, okay, this needs to translate into this to get the proper output. Let’s talk about kind of future of work. Everyone says A is going to redefine the word, but what actually does that mean for what you think it means for a superintendent or a PM or an administrator or somebody like that? What does this do for these folks? an administrator or somebody like that.
(34:44) What does this do for these folks? One thing it does is there’s always an assistant available to help you, you know, no matter the date, the time, or the venue, as long as you have a good internet connection. I mean, they have some edge AIs too but they’re not probably as good as the audio AI personally for me when I think about I use you know these models so often Gemini GPT the other day you know this weekend actually I was working on a voice to text model for construction, noise and construction kind of text.
(35:28) Like, let’s say you are in a field and you’re doing some sort of, let’s say, daily reporting at the end of the day. So this model, you know, throughout the day, you talk to it in voice and at the end of the day, it just creates the daily log, right? And to build that kind of model, you know, somebody like me, a couple of years ago, it would have taken me a month, you know, and, you know, to get to a good starting point, it just took me a couple of days, you know, I just started with the Whisperer
(35:59) and then used the coding AI agent with Gemini and I started asking, give me the synthetic data for a construction job site so that I can start training this voice-to-text model. It gave me the code, and then it gave me the way to test it, and it gave me a strategy how we can go from here, build an MVP, and get a fully functional construction ASR, right? So the progress, in my view, that has been really significant is in coding, you know.
(36:43) Actually, software programming has been disrupted hugely in the last two, three years because these AI models are very good at coding. Yeah. So I hear the same things so i hear the same things i hear the same things and it does make me again you know we were talking earlier about how companies the consigli um topic you know well does a does aecom want to be um a construction company are they trying to be a technology company are they are they you know trying to pull some vertical out of technology that you know they specifically use i
(37:12) don’t know what they’re trying to do but again this framework this kind of example i guess and leads other people to believe you know well it’s so, well, it’s so easy for you. It sounds so easy for you to develop these models and, you know, develop my own software. Why don’t I just do it in-house? And then we lose focus, right? Because the things that I worry about probably aren’t the turners of the world or the AECONs of the world kind of folks.
(37:44) I’m not worried about them. They already have departments that are, you know, 12 times larger than any company I’ve probably worked for or my own company when I ran it. But what I’m thinking about is these mid-sized contractors who at some point are probably going to need some sort of digital proof that they are competent and skilled in providing what’s needed meaning they’ve got a digital track record of their safety they’ve got a digital track record of paying their bills on time they’ve got a digital track record
(38:26) of paying their bills on time they’ve got a digital track record and i’m saying it’s it’s a performance track record and i’m you not only are we as construction personnel and construction people getting educated on the on the benefits of digitization but clients um and people who are you know buying product buying buying buildings, building things, doing those kinds of things, they’re also getting very well educated on what’s possible.
(38:52) And I don’t see it being too long before you’ve got to come to the table with, I’ll call it a digital receipt of your performance. That has an impact on not only contracts it has an impact on pre-construction and has an impact on construction scheduling it has an impact on closeout everywhere across the board but you’ll have to provide some sort of you know digital proof of concept or or proof of performance what’s your thoughts on that? Yeah, and one thing I think is changing really fast is the amount of digital information
(39:34) we are acquiring and construction at the same time, it tends to lag behind almost every technology wave, right? So, I mean, there’s a technology called blockchain, which can do some of the things that you’re talking about very well, right? I’m not sure whether you were alluding to that or not. I mean, I believe to realize, and I’m sure you are talking to people like that too, that change is happening, you you know and we as society have progressed so much over the last you know 100 200 years because we were willing to adopt
(40:15) the change right i want to go back to that point you’re making about like a construction company acquiring let’s say an ai company right uh you know one fundamental tenant of our society’s big success has been division of labor right i mean if i’m good at coding and you are good at let’s say building buildings i mean why don’t we just focus on what we are good at and we exchange, right? Instead of you and I trying to do the same thing, maybe I acquire your company and try to start building buildings too, you know, and I kind of dilute my focus.
(40:57) There’s 6 billion of us in this world, right? And if everyone starts do we do a lot of things suddenly the quality of product that’s going to be out in the market is not going to be high enough yeah specialization being proficient at your craft whatever that is um i do worry about that becoming about that becoming maybe as you diluted right because what is it sometimes enough information will make you dangerous you know it doesn’t make you proficient but it makes you it makes you dangerous because you know enough about it to maybe perform in the space but you don’t know
(41:40) enough about it to understand what the outcome means. Yeah, I think I think that’s important as we as we are humans. Right. And I think that gets back to your even the just the title of your your whole company, you know, the name of your company beiing human. We’ve got to keep the human element involved in this and still understand, you know, we need this.
(42:03) Right. We need not only folks that understand the use of ai but we still need people that upskill at plumbing welding you know all of those things we need those specialties because i guarantee there’s not a tech guy out there that’s going to be able to then jump out from behind the computer to go you know weld a nice bead on an oil pipeline.
(42:27) Yeah, that’s a different skill set. And the tech guy is counting on the plumber to be able to do that, right? And the plumber is counting on the tech guy to be able to create good software, right? Yeah. Well, all of these things concern me when I start to think about construction because I do think about the some 500,000 jobs that will go unfilled.
(42:52) It was this year. It was the same number it was last year as it was the year before. We’re rapidly losing what I’ll call tribal knowledge about certain things, important things, electricians, plumbers, welders. And these aren’t the folks that are going to be utilizing what we think is AI. But I get a little nervous when I hear the pundits start to talk about how AI is going to help us transform maybe a labor shortage or anything like that and i don’t see it i i see where it can save time um for doing redundant
(43:34) activities you know maybe within the office structure or you know those types of things getting them so they’re not waiting on something and keeping them busy that’s a productivity standard but that’s not a that’s not a scalable you know um people standard uh so i do worry about that i think uh one important thing is the physical ai that revolution is coming to next 10 years you know right now in san franc a mo and Tesla they are having driverless cars running everywhere so there’s a whole plethora of people who
(44:11) are relying on being able to drive the cars and make their livings their jobs gonna go away in next five to ten years because you know that technology is coming it’s hard to say that what can be automated and what cannot be automated. Also, you don’t see people being able to automate like a plumbing job, you know, maybe a robot doing plumbing work.
(44:37) I can’t imagine that happening anytime soon, you know. But again, we really don’t know how far this is going to go in next you know when however many years 20 25 years yeah well i think that always begs the question and it’s a good one um is i can’t remember who said it but it’s not about what you can invent but whether you should invent it is always the question for science in itself, right? Obviously, sky’s the limit on what can be invented.
(45:14) But the question is, for the sake of human race or the sake of the humankind, is should we do it, right? Should we be trying, should these efforts, you know, be traveled? You know, one thing I want to ask you before we kind of wrap up here is, is you got what’s one thing right now every construction leader should do probably this month to prepare for for AI without spending a dime? for AI without spending a dime.
(45:48) Mike, what should they be doing this month? A good piece of advice for just this month or maybe Q1. Listening to quality information, including your podcast, right? We’re talking about great stuff here. A lot of times we are even construction leaders, underestimate the importance of education. Yes, I agree with that. Very good. That’s good advice.
(46:14) I come from the Warren Buffett school of try to read something, read as much as you possibly can, get as much of an education as you can. And I’m not talking about, you know, asking a question to chat GPT either. I’m actually talking about getting out there, maybe getting a book in your hands or getting a book on tape, listen to some podcasts, get some different point of views going, really educate yourself on what’s out there.
(46:41) I know there’s not a lot of time in the day, but there is, you know, maybe there’s drive time. Maybe there’s, you know, 15 minutes a night. It doesn’t take that much, but it does incrementally increase to us for us to educate ourselves on these things. So I think that’s great advice. Great advice. Rishi, it’s been fantastic.
(46:59) You know, time flies when you’re having fun and you’re, you know, chatting about these things. I really appreciate your insights. when you’re having fun and you’re you know chatting about these things i really appreciate your insights and um how can folks get a hold of you and uh and and get a hold of beiing human we are active on linkedin so the company’s name is beiing human b-e-i-i-n-g space h-u-m-a-n H-U-M-A-N our website is beiinghuman.
(47:26) com again being with two I’s my email is rishi at beiinghuman.com R-I-S-H-I at B-I-I-N-G H-U-M-A-N.com yeah those are the channels very good well folks I will put all that information into the into the show notes we’re going to be publishing this obviously on spot on spotify for your drive time listeners uh if you like the video versions of these things um i also have it on my youtube channel um but rishi thanks so much for joining me this week i appreciate it thanks again Thanks for having me on. Bye-bye.