The AI Age for Fuel and C-Stores. Vlad Collak, Co-Founder and CEO of NewTide.ai, joins Doug Haugh Chairman
Greatest Disruption in History
Vlad Collak, NewTide.ai Co-Founder and CEO, shares the key to thriving in the greatest disruption ever to hit fuel and c-stores.
AI is a once-in-a-generation opportunity to create intelligent processes and systems that solve real problems and establish a competitive advantage for the future.
You'll discover the specific use cases to reap immediate benefits.
And, how to reduce your risks by aligning PROVEN industry partners doing the real work in Fuel and C-Stores.
You'll Also Discover:
A First Step to Get Started.
The Proven Strategy for Integrating AI.
Why AI is Not "Rip and Replace".
The Unexpected Benefits.
The Key to Scaling Your Business.
What are your thoughts on the future of AI in the energy and
convenience industries?
--------
Follow Doug Haugh on Linkedin: https://www.linkedin.com/in/douglashaugh/
Follow Vlad Collak on LinkedIn: https://www.linkedin.com/in/vladimircollak/
--------
This episode is powered by NewTide.ai, Enterprise AI Built for the Fuels and Convenience Industry. Learn more here: https://newtide.ai/
Transcript
I think it comes down to knowing the questions to ask.
Speaker A:If you don't have the experience in the business to know where the important questions lie, you're starting out behind.
Speaker A:There's a lot of nuance in our business to figure out customers are really.
Speaker B:Not served as well as they could be, and that's because they're using traditional software.
Speaker B:Five years, 10 years from now, it's going to be all AI.
Speaker B:Because of our industry knowledge, our passion, we can actually provide value to customers a lot quicker because we get their use cases.
Speaker C:AI is the next big boom for fuel and sea stores.
Speaker C:You know you need it because you're drowning in data, yet sometimes lack insights.
Speaker C:You're stretched thin and the AI learning curve can feel impossible.
Speaker C:Your competitors are starting to move.
Speaker C:Are you getting left behind?
Speaker C:Well, welcome back to Fueling AI, the podcast that cuts through the noise and delivers the real story on AI for fuel and C stores.
Speaker C:Hosted by industry veteran and CEO Doug Hawke, this is your no nonsense guide to safely leveraging AI to drive profitability and growth.
Speaker C:We'll expose the risk, explore the possibilities, and give you actionable strategies you need.
Speaker C:Now subscribe and ignite your AI advantage.
Speaker A:Welcome to Fueling AI.
Speaker A:I'm your host, Doug Hawk and joining me today is Vlad Kolak.
Speaker A:We're going to talk about latest trends, technology, team building, all kinds of topics around AI today.
Speaker A:And Vlad, thanks for joining us.
Speaker A:So excited to have you.
Speaker B:Yeah, thanks for having me.
Speaker B:I'm looking forward to this.
Speaker A:So I'm going to start with why.
Speaker A:I mean, there's obviously a ton of hype across the industry and we read about it in the press every day.
Speaker A:We can talk a little bit about our backgrounds.
Speaker A:We've known each other a long time, I think going on probably 25 plus years now, worked on all kinds of technologies, were early pioneers in the dot com days, I think, and bringing the Internet into our industries.
Speaker A:Why do you think now is the right time for AI in particular in our business in fuels and energy and convenience retail?
Speaker B:Well, you touched a little bit on my background, but many people probably listening don't know that you actually hired me at Fuel Quiz, one of your early startups, 25 years ago I was like an employee number five engineer, just doing all the engineering work and like I had no clue about the business.
Speaker B:Right?
Speaker B:Like I didn't, I wasn't exposed to fuels, I wasn't exposed to convenience stores and.
Speaker B:But I saw quickly there was just this complexity in the business which was really interesting.
Speaker B:Right?
Speaker B:Like a lot of problems to solve.
Speaker B: And this was early: Speaker B:So couple that with, okay, now we can do it in a new way.
Speaker B:There's this Internet thing, right?
Speaker B:And we can solve these problems in a way that have never been solved before.
Speaker B:So that was really, really interesting.
Speaker B:And you know, we always talk about this.
Speaker B:This was so much fun building new things.
Speaker B:We were building cloud solutions really before there was Word cloud, before AWS ever existed.
Speaker B:We were connecting ATGS to the Internet, which is now IoT.
Speaker B:We were automating infrastructure where you can plug in a server, spins up, installs the operating system, puts on the database and it just goes.
Speaker B:We all felt like we were innovating during field quest times and I think now is that same way.
Speaker B:Right.
Speaker B:The AI wave is really the once in a generation in my view.
Speaker B:And I know you feel the same way that we can create now intelligence.
Speaker B:It's not just software, but it's intelligent software solving real problems.
Speaker B:I love this industry.
Speaker B:I've been in it now for 25 years without even trying to begin with.
Speaker B:And I appreciate the importance this industry has, whether that's energy and that energy is running the world.
Speaker B:We can't do anything about energy, obviously, and the convenience where Sea store retailers are kind of the bedrock of many communities and I find that appealing and I want to help them take their businesses to the next level, which there's no better way than using this new set of tech.
Speaker A:Yeah, I agree.
Speaker A:I mean, I think we've got a tremendous opportunity in front of us.
Speaker A:I think that it is.
Speaker A:I think timing is right for most businesses to get started and we'll talk a little bit about sort of the paths forward.
Speaker A:But I do believe it's time to get started and get going.
Speaker A:Right, and take those first baby steps, get your first solutions deployed, get your employees exposed to.
Speaker A:What does this really mean at work?
Speaker A: It does feel like: Speaker A:We started searching and I mean it was.
Speaker A:It was like, oh, this is super useful, but how the heck am I going to use it at work?
Speaker A:Like, how do I integrate it into my work environment?
Speaker A:How do I managed security and privacy and all those things that we wrestled with.
Speaker A:I see all those questions resurfacing now in a different context, but with the same drivers that people are using this.
Speaker A:You're using chat, GPT or Gemini or cloud or one of them at home or on their phone and they're like, darn, this is pretty useful.
Speaker A:Like, I Mean for all kinds of things, Right.
Speaker A:But how do I get it integrated in the enterprise?
Speaker A:So maybe along those lines, like what are some simple examples?
Speaker A:Or not even, not even necessarily use cases, but just ways to, for people to think about.
Speaker A:Where do I start and how does this, how is this relevant to me if I'm a fuel distributor, fuel supplier, convenience retailer, what does it mean?
Speaker A:Because I'm not, I'm not chasing the chat GPT model wars.
Speaker A:Like I just want to automate something in my business that makes me money.
Speaker B:Yeah.
Speaker B:So I think the way I think about this is it's still super early.
Speaker B:Back to the Internet days.
Speaker B:We're like, well, how do we apply this technology in an enterprise context?
Speaker B:One, it's just getting started.
Speaker B:This will evolve super quickly, but it takes time for it to really make it all the way to all of the parts of the economy and every job and every company.
Speaker B:I would start with thinking about, okay, great.
Speaker B:So what tools do we have today?
Speaker B:Many of us use Google for example, or use something like SharePoint.
Speaker B:We have documents and most people, there's no shortage of data.
Speaker B:But how do you find insights?
Speaker B:Can we use these AI tools to really get faster at the insight?
Speaker B:The analogy I use is, yeah, I can go Google something or search something, get the results and I can get a bunch of articles and then I have to read through every article, read them really well, figure things out.
Speaker B:It takes time.
Speaker B:Well, AI now can read really well, right?
Speaker B:And so we can ground our AI enterprise AI systems with the industry knowledge, with knowledge of the organization.
Speaker B:Now all of a sudden, employees can just have this assistant, right?
Speaker B:Could be as simple as a.
Speaker B:Think of it as Chad GPT.
Speaker B:But one for you, that's yours, that's private, that's secure and grounded to your organizational data.
Speaker B:So I think that's one way to get started.
Speaker B:And then it's just really expanding that to every role.
Speaker B:So with these AI tools, right, they're only as good as one the data you give it and kind of the instructions, Right.
Speaker B:People probably have heard of this thing called prompt engineering where you can ask a simple question, you get back a decent answer, but if you really think about the question, you can really get a lot better results.
Speaker B:And so that's where I find this appealing.
Speaker B:Because if you're within the industry, you know what to ask because you have specific things.
Speaker B:You can train these models to be really, really specific.
Speaker B:And so at New Tide, that's our job.
Speaker B:It's just, how do we build a platform that is grounded a company data Organizational data of our customers, but very tailored to certain roles.
Speaker B:So I'm like a trading analyst or I'm credit manager or dispatcher.
Speaker B:It knows how to do my job, right?
Speaker B:And that's kind of.
Speaker B:So that to me, start small, start simple, put your data in one place and let AI use it.
Speaker B:And then you'll probably find a bunch of different things.
Speaker B:That's one sort of case.
Speaker B:The second one is there's all these manual processes, right, throughout every organization.
Speaker B:And so I really think about this as you look at the evolution of technology.
Speaker B:It was.
Speaker B:Everything was on a piece of paper, right?
Speaker B:And then we're like, okay, let's digitize this, right?
Speaker B:So now we have accounting systems, ERP systems.
Speaker B:But if you look at like an invoice, it's still typically like a PDF invoice, right?
Speaker B:That system sends an invoice to a customer, some human reads it, have to figure out what to do with it, and then feed it into another system.
Speaker B:So you got these two systems in the edge and human in the middle for literally no reason, right?
Speaker B:And so now we can have AI in the middle.
Speaker B:So there's this technology integration end to end, which actually streamlines the process because we don't have to have an army of analysts or army of data entry people.
Speaker B:And those employees can now focus on really a lot more important tasks.
Speaker B:How do you grow the business?
Speaker B:How do you take care of the customer and that kind of thing, you know?
Speaker A:Yeah, no, I think that's critical.
Speaker A:I think, I mean, look, we've been automating transactions for, for many, many years, but it's, it's only structured transactions.
Speaker A:And I mean, how many customers can you actually connect to your systems live?
Speaker A:And then how much work is that?
Speaker A:Where now you basically have this extremely flexible layer in the middle.
Speaker A:It can read like we do, right?
Speaker A:It can understand the context and it knows the relationship you have with that customer and it can reference the contract that's in place to service that customer.
Speaker A:So if there's questions on the invoice, the customer doesn't have to call back.
Speaker A:I mean, they can just interrogate it live, right?
Speaker A:So I think it's, We've always, we've always had this data and over the past, I mean, you and I have worked on this for decades now of streamlining those data flows, putting them in the structured pipes, then program the APIs and connect them to the other systems.
Speaker A:And where we do that, there's great returns, but you can only do that for high volume, like highly structured situations.
Speaker A:And they Just aren't very flexible.
Speaker A:Right?
Speaker B:Yeah, they're pretty brittle.
Speaker B:Right.
Speaker B:Like basically something changes all of a sudden something breaks.
Speaker B:Right.
Speaker B:Versus if you use something like an AI in the middle.
Speaker B:Right.
Speaker B:We'll talk about large language models.
Speaker B:I'm sure later that thing can adjust to whatever it's seeing.
Speaker B:Right.
Speaker B:Almost as a human would versus something that traditional software can do.
Speaker B:You program it a certain way.
Speaker A:Right.
Speaker B:Some if statements and some sort of corner cases.
Speaker B:But by and large it's not thinking.
Speaker B:Right.
Speaker B:Versus these technologies can.
Speaker B:And then you kind of hit on something very, I think subtle and important though is that sure, we could integrate, say we have this whole supply chain.
Speaker B:We can integrated into and between all these different systems through something like APIs.
Speaker B:Well, one that's expensive, right.
Speaker B:If you wanted to do something like, okay, let me get, let me get truck tech integration for something like Skypitz.
Speaker B:Right.
Speaker B:And I want to be able to get the delivery tickets coming in.
Speaker B:Well, they don't have.
Speaker B:They have large market share and I don't know what they are.
Speaker B:I think they're like 10% or something of the market.
Speaker B:Maybe somebody can correct me on that.
Speaker B:But.
Speaker B:So then I integrate with them.
Speaker B:Now I'm going to have to integrate with someone else and someone else and I'm as IT team, I'm just chasing my tail all the time.
Speaker B:So now you can say no, just email the delivery tickets to me, my AI will read them and then do something with them.
Speaker B:And I don't care what system it is because it knows this delivery ticket from that one in a subtle way.
Speaker B:Now integration becomes super easy because AI is reading that and kind of adjusting based on what it sees.
Speaker C:Yeah.
Speaker A:I think a lot of those cases where we would in the past probably body shop it.
Speaker A:Right.
Speaker A:And you just.
Speaker A:Some of those have been offshored where we've got floors full of people literally reading documents and entering them in systems.
Speaker A:Right.
Speaker A:I think those things now can be easily automated.
Speaker B:Yeah.
Speaker A:Well, tell me why.
Speaker A:You know, why New Tide?
Speaker A:Why.
Speaker A:What's the.
Speaker A:What's the specific mission?
Speaker A:There's, there's people throwing tens and hundreds of billions of dollars at this.
Speaker A:Where does New Tide fit?
Speaker A:Why does the customer need you?
Speaker A:Like, what's the, what's the niche or purpose that you see filling in terms of bringing the company to market?
Speaker B:Yeah, yeah.
Speaker B:So I already talked a little bit about sort of this nexus of energy and convenience.
Speaker B:Right.
Speaker B:The things that we've been doing for decades now and then the new technology and I think that's really the critical piece is that we feel like today a lot of those sort of customers are really not served as well as they could be.
Speaker B:And that's because they're using traditional software, right?
Speaker B:So if we look at sort of five years, 10 years from now going to be all AI and I think they can get a lot more sort of value if we start leveraging those technologies.
Speaker B:But why you tied.
Speaker B:You know, we, I mean, there could be probably some Silicon Valley startups coming in a space and building solutions.
Speaker B:We feel like because of our industry knowledge, our passion, our sort of understanding of the whole space, we can, you know, this solutions could be coming from the industry, not from some outside thing where they're trying to figure out the market, trying to figure out the processes, right?
Speaker B:And now it's taking forever to get value.
Speaker B:So we think we can actually provide value to customers a lot quicker because we get their use cases right.
Speaker B:Many of us with a new tide.
Speaker B:You in particular, right?
Speaker B:You guys ran much of these businesses that you were in those customer shoes.
Speaker B:And so I think that's kind of the important aspect of that where we're connected to the industry as opposed to just some take startup coming in, trying to figure things out, you know.
Speaker A:Yeah, I think, I think that's right.
Speaker A:I think, I think we're going to see a host of verticalized solution providers that can accelerate the technology, accelerate the adoption.
Speaker A:I think it comes down to knowing the questions to ask, right?
Speaker A:These, these systems now are very good at giving you answers and finding the answer.
Speaker A:But if you don't have the experience in the business to know where the important questions lie, you know, you're starting out behind.
Speaker A:And I think, I think the other thing is that sort of understanding the difference between a barrel and a gallon and you know, a tank wagon versus a tanker truck.
Speaker A:And it's just, there's a lot of nuance in our business that I think is going to take a generic solution a long time to figure out.
Speaker A:From a, from a technology perspective, does that, how do you take that experience, passion, knowledge and impart that into the.
Speaker A:How does that technically translate to an acceleration of what the platform can do, what your agents can do, what, what level of understanding they show up with, versus jumping out of the box in a generic sense.
Speaker B:So certainly if you look at sort of building a company, building a solution sense, right?
Speaker B:If you don't have a good, what they call founder, market fit, right?
Speaker B:Founders come in, in the market, they don't know anything.
Speaker B:There's this large steep learning curve, right?
Speaker B:So there's this delay of like, we're gonna have to spend next 12 months just figuring out what everything is, talking to customers until we sort of get it right.
Speaker B:So that goes away, right?
Speaker B:Like that, we have that fit.
Speaker B:We know, we, we, we know where the problems are.
Speaker B:Now we just need to get the customers, tell us what is kind of the current priority for them.
Speaker B:Right?
Speaker B:But the second thing is these AIs, while they're amazing, right, and they keep getting better and better, they're only good as the data, right?
Speaker B:And so typically These look at ChatGPT or all these copilots, well, they're grounded, but they're grounded in the data that's on the Internet, Right?
Speaker B:Anything that's on the Internet, they know.
Speaker B:But they don't know your organizational data.
Speaker B:Right?
Speaker B:Which means that when it comes to specific trading or dispatch or even running a C store and store operations or category management, they don't know that.
Speaker B:So it's our job and our ability with new tie, to bring the data in.
Speaker B:And the way we look at it is we first ground these AIs with industry data, right?
Speaker B:What are the terminals?
Speaker B:What are the different suppliers of those terminals?
Speaker B:Industry terms, right?
Speaker B:Anything that kind of maps out the supply chain, the lingo of it all.
Speaker B:And so our AI comes with it, right?
Speaker B:So we don't have to start from, we know we're starting from a lot higher base, if you will, because.
Speaker B:And then we layer in the specifics of the organizational data.
Speaker B:And you know, I think the analogy we're using now is, which is, I think it's a really great analogy.
Speaker B:It's like, well, think of these things as your digital employee.
Speaker B:Okay, great.
Speaker B:Well, do I want to hire like a college grad, right?
Speaker B:Or do I want to hire someone that's 20 years experienced in the industry and say the price is the same or even cheaper, Right?
Speaker B:Like, why not hire the, the digital employee who actually knows more than, than the college graduate, right?
Speaker B:So that's, that's that knowledge, that institution, that organ industry knowledge that I think is important to have, you know.
Speaker A:Right.
Speaker A:Well, speaking to that, like, where's the line?
Speaker A:How do customers think about what's theirs and what they own?
Speaker A:And, and you've heard me talk about this.
Speaker A:I think it's, I think all enterprises are faced with, you know, a complex set of choices going forward where, you know, how do I transform this data into a true asset that can be used to drive automation, customer service, sales, marketing activities?
Speaker A:Because it understands my institutional knowledge as a company and it understands my competitive advantage, like how I win.
Speaker A:So how do I, how do I do that?
Speaker A:Create an asset that leverages all that and not give it away to different software company, service provider, et cetera.
Speaker A:That then essentially I've taught someone else's employee to be very smart.
Speaker B:Yeah, yeah.
Speaker B:I think it's making sure that the incentives are aligned.
Speaker B:So, for example, we're not model builders at New Tide.
Speaker B:Right.
Speaker B:We're leveraging existing models that could be open source, could be commercial.
Speaker B:We actually pick the right model for the right job and really orchestrate it all in a workflow or solution that is ready for the customer.
Speaker B:Right.
Speaker B:Well, others.
Speaker B:Right.
Speaker B:If you're a research lab, you might have an incentive to like, I need to improve my model.
Speaker B:I need to train this model based on data.
Speaker B:And your incentives are not actually aligned not to do that.
Speaker B:Right.
Speaker B:And so we're very careful to make sure that, look, you know, this is your data.
Speaker B:Right.
Speaker B:If we bring in industry data into our AI, it comes with it and it's public data.
Speaker B:Right.
Speaker B:Something that anybody could download when we populate guys with that.
Speaker B:But then when customers bring in data, we actually look at it from like an enterprise standpoint.
Speaker B:So just like if you upload your data to SharePoint, well, you're not going to go share it with the rest of the world and we're not going to do that.
Speaker B:Right.
Speaker B:On your behalf.
Speaker B:It's the same thing where it's your data, it's your environment, you can continue improving these agents.
Speaker B:And that's kind of.
Speaker B:You mentioned bringing your.
Speaker B:Sort of building your data assets.
Speaker B:Right.
Speaker B:Like these agents kind of become yours because it's trained on your data, your specific culture processes, the way of doing things.
Speaker B:Right.
Speaker B:And so now actually they're not all that useful to everybody else because everybody else might be.
Speaker B:Might be doing something differently.
Speaker A:Right.
Speaker B:So that's also kind of aligning incentives.
Speaker B:We don't need to take these agents and share them everywhere else.
Speaker B:We want the companies to have the best agent, which means highly personalized on their data, you know.
Speaker A:Right, right.
Speaker A:I do think we'll have a.
Speaker A:I think it's helping customers and folks in our business adapt and adopt the different pathways.
Speaker A:Right.
Speaker A:Like, I mean, if you're on SAP for ERP, or you're on Salesforce for CRM, or you're on TMW for dispatch, or you're using Gravitate for truck scheduling and tank monitoring or something like that, or Titan's got all kinds of talk about a massive data stor.
Speaker A:I think they have 30, 40,000 sites that they're monitoring every day.
Speaker A:They're going to have embedded enhanced capabilities that use AI.
Speaker A:Right.
Speaker A:I think it's important for folks to understand like that's fine, that's well and good and I want to push those service providers to do better work for me and get more out of it.
Speaker A:But I think what you're talking about is those are embedded in those solutions that sort of use that industry software to get something done that's different than building your agent that then uses it in your specific.
Speaker B:Yeah, because you use obviously there's a bunch of point systems, right.
Speaker B:And that's great.
Speaker B:And they should use AI to make it sort of experience better.
Speaker B:But then I need to know my entire business end to end.
Speaker B:Right.
Speaker B:So what is happening with my supply and trading versus what's happening with my dispatch?
Speaker B:And that data sits in two different systems, right.
Speaker B:So if I can bring it in in one place and have sort of one pane of glass on top of each wei that I can, I can get all kinds of insights that I never would have gotten otherwise.
Speaker B:And that's really what we're talking about.
Speaker B:It's really no different than having most of our customers have these analytics packages.
Speaker B:They take all this data from all these different places, build out a data lake, put analytics on top of it, but then human has to go look to see what's happening.
Speaker B:Here's a chart.
Speaker B:Let me see.
Speaker B:They spend hours trying to figure things out.
Speaker B:That's where you can use that base layer, the components layered together.
Speaker B:Right.
Speaker B:With data lakes, with bringing all the data together, but then put a layer on top of that where we have a lot more natural sort of interaction with, with the tool, but also insights that you maybe not have gotten.
Speaker B:Because AI can read thousands of pages of documents and reams of data tables right in split second versus humans just can't.
Speaker B:Right.
Speaker C:Feeling overwhelmed by the AI landscape.
Speaker C:You're not alone.
Speaker C:Most Fuel and C Store leaders know they need AI.
Speaker C:But figuring out where to start can feel impossible.
Speaker C:That's where Doug's team at New Tide is offering an opportunity for a 20 minute AI blueprint consultation.
Speaker C:During your session, you'll work directly with Doug's team to map out a personalized strategy for your business.
Speaker C:They'll pinpoint the areas where AI can deliver the biggest impact, from optimizing inventory and product pricing to enhancing customer service and boosting security.
Speaker C:They'll also discuss the potential risk and how to mitigate them, ensuring a smooth and successful AI integration.
Speaker C:Go to www.
Speaker C:Newtide.AI and sign up for your free 20 minute AI Blueprint consultation today.
Speaker C:That's www.
Speaker C:Newtide.AI.
Speaker C:don't wait, your business depends on it.
Speaker A:What about, like, how does you know if they're making those investments?
Speaker A:They, they've done the work to get their data into a system, they've created those assets.
Speaker A:How do they know who the winner is going to be?
Speaker A:Or do they have to?
Speaker A:I mean we just had a, we saw a 600 billion dollar disruption last week or week before with this deep seq release.
Speaker A:Do, do the users, do people have to pick a winner?
Speaker A:Do they have to have a prediction as to who's going to win?
Speaker A:Like, how do you, how do you leverage all of that activity that's going on?
Speaker A:And yet while you're continuing to do your work to advance your assets and your capabilities and quote your agents, how do I not like build in something that's obsolete and 90 days at the fast, as fast as this stuff's moving.
Speaker B:This is moving so incredibly fast, right?
Speaker B:Everybody, month ago, OpenAI was the king and then Deepsea comes out and they have a model that's presumably just as good and a lot cheaper, right?
Speaker B:This is just the beginning.
Speaker B:This is absolutely arms race and we can see big tech spending billions and billions of dollars.
Speaker B:They're talking about trillions now, right, to work on these models.
Speaker B:So the reality is going to be, is just it's going to be impossible to pick a winner.
Speaker B:And then sort of the dark side of that is if you do pick one of those, then if you're very specific to that vendor, then you may not take advantage of all the sort of advances somewhere else.
Speaker B:Eventually they all sort of catch up, right?
Speaker B:So kind of maybe washes itself out.
Speaker B:But the way we look at it is like, well, no, pick a industry specific vendor that knows your industry provides those solutions, but make sure that vendor is agnostic, right?
Speaker B:So doesn't care, hey, do I use Deep SEQ or do you use OpenAI?
Speaker B:Do you use Meta's?
Speaker B:You know, llama?
Speaker B:Well, actually in reality some models are better than others, right?
Speaker B:For certain tasks.
Speaker B:So they, everybody talks about Sonnet from Anthropic being like the best coding assistant, right?
Speaker B:Versus you know, Deep seeks kind of this lightweight sort of reasoning model that, that seems like it's actually doing pretty well and it's not, it's not sort of as heavy in terms of compute infrastructure.
Speaker B:So I think in reality there's going to be bunch of advancements and at least from our perspective from New Tide, we'll take advantage of all those, right.
Speaker B:We'll, we'll watch the trends, we'll pick the right vendors for the right job, but be always grounded in what the customer is actually trying to accomplish.
Speaker B:Because forget all this tech stuff, like at the end of the day we're trying to get our job done, trying to grow our business.
Speaker B:And so we started with kind of what are the objectives and then figure out, okay, how do we, how do we get there using the best tools, you know, so I wouldn't at this point, I like, it's anybody's guess who's going to be the winner, you know, so it's just hard to pick one.
Speaker A:Yeah, well, and just give us a little bit of, let's not go too deep down the rabbit hole in tech, but if one wanted to use a model like Deep Seq as an example, which there's obviously massive security and privacy concerns, potential Chinese government sitting behind it or whatever the pipelines might be, you know what, how do you create the environment and integrate the latest tools into the platform, provide those capabilities to the customer while not exposing again back to data security and privacy and protecting their assets and yet being able to truly access if there is a breakthrough.
Speaker A:But I don't like kind of the hosted model.
Speaker A:Are there other ways to do that?
Speaker B:Yeah.
Speaker B:So I guess first of all, please do not call the deep seq APIs or open up their UI.
Speaker B:Right.
Speaker B:Like I guarantee that data is leaking into the Chinese government.
Speaker B:Right.
Speaker B:So that's not the right answer.
Speaker B:But Deep SEQ also has a open source model.
Speaker B:It's completely open code, open way.
Speaker B:It's not to get too technical, you can literally use it and it could be in your infrastructure.
Speaker B:So whether you're literally on prem, running on your servers, can never reach the Internet or in some cloud provider like Azure and AWS and Google, which they all have deep, I think at least Azure, which we're on, they have Deep SEQ and I think others as well.
Speaker B:Right.
Speaker B:So they, what we do is basically one, do it in a agnostic way where if we feel like for that particular job, OpenAI APIs are the right thing.
Speaker B:We actually go through the cloud provider, right.
Speaker B:Not directly through OpenAI because even Microsoft and all these other big cloud providers, they, they shield you from liability, they shield you from sort of, not that they have Secure sort of SoC2 hip, all this stuff compliant infrastructure.
Speaker B:So us as a software provider, we clearly leverage those, but then we're able to deploy open source models.
Speaker B:Right.
Speaker B:So even this morning we talked in our engineering call about like hey, what model do you want to, want to use for this particular task?
Speaker B:And there's different options and some are, yeah, let's grab an open source model because it's good, it's inexpensive, we can host it, we control it, and then for this particular task use that.
Speaker B:Right.
Speaker B:While in some cases we call some API like Azure OpenAI or you can do Gemini, which is Google product or aws.
Speaker B:Right.
Speaker B:So, so there's a little bit of that cloud infrastructure that we sit on top of, which actually helps us being agnostic, you know.
Speaker A:Yeah, I thought the, I thought the Deep SEQ example was important for people to see just because I think and for, for those listening, Vlad showed me this a week ago.
Speaker A:I think he deployed it on his laptop, which was stunning.
Speaker A:Stunning to me that it could run on a laptop with not a lot of computing power and it was answering pretty complicated questions around.
Speaker A:I was making up trades and firing different potential, hey, a barge move on the Mississippi from Valero to Exxon or this.
Speaker A:And it immediately understood what I was talking about.
Speaker A:It deciphered that into data, converted it into a format that could be loaded into a system, and it did it on your laptop.
Speaker A:Which what got me excited about that is I've worked in refineries as well over the years where our systems, particularly any, anything that could potentially be touching controls are air gapped to any network and it doesn't matter if there's firewalls and there's different network mappings and so forth, that's not good enough.
Speaker A:Like it has to run completely independently.
Speaker A:And that was the first example that personally experienced where I'm like, oh, I could actually start to deploy standalone on prem models that can be air gapped, not connected to anything other than the data I want them working on and still bring that, still bring a pretty high level of intelligence to the situation, even with, in that case, very, very modest computing power.
Speaker A:Right.
Speaker A:So I think that was, that was.
Speaker A:I know we've run llama tests and similar, but didn't show quite the same capabilities with that very limited computing capability.
Speaker B:Yeah, yeah, no, I think that's a, that's really one of the reasons, I think a lot of people got excited because now you can start pushing this stuff more to an edge.
Speaker B:Right.
Speaker B:So you'll see more of these examples where these models could run on your phone, could run on your embedded controller and still be kind of safe and secure.
Speaker B:Not connected to especially like a control system.
Speaker B:Right.
Speaker B:Connected to the web.
Speaker B:I think for us is back to, hey, let's Solve problems in a way that makes sense.
Speaker B:And some of that is well, we're not beholden to any one technology.
Speaker B:Right.
Speaker B:So those customers who use us is this thing underneath shifts and moves and evolves.
Speaker B:Right?
Speaker B:Right.
Speaker B:Stable layer with which they integrate with and underneath it we can go do things.
Speaker B:It's like, like if it's, it's like being able to work on an engine while you're driving down the road, you're still driving, you don't know anything.
Speaker B:And then underneath it we're tweaking the engine and improving the performance while you're on the road.
Speaker B:Right.
Speaker B:So that's kind of the analogy I think new tight can bring to, to customers.
Speaker A:What do you think about.
Speaker A:You mentioned something earlier about a single pane of glass and I, I think of like the trade floors that I've worked on where there's like an ever scaling monitor rack around the best traders.
Speaker A:Right.
Speaker A:Where or at least the most demanding traders who can get the infrastructure.
Speaker A:But I mean there's, there's one on.
Speaker A:I was on a floor last week and it, it was, it was like half encircled globe around the guy and I don't know how many screens it ended up being.
Speaker A:Maybe it was 15 plus.
Speaker A:Doesn't this sort of provide a different way to do that?
Speaker A:Where the reason, the reason traders do that and I've done it in the past is there's certain insights and situations and data feeds and yeah, there's arbs, there's spreads, there's price points, there's, there's physical flow of data and volume and inventories that I want to see real time as much as possible.
Speaker A:There's other things that drive value in our, in particularly of those of us in the energy business because weather, I mean obviously it has a big impact if you're in a construction business or one of those but for us it can really change the demand pattern of a day, of a week.
Speaker A:It obviously impacts your physical ability to move things but more than that it impacts how many customers might show up today.
Speaker A:But the traders today are trying to synthesize all of that by exposing it on a screen, putting a bi layer, creating a new chart, a new graph.
Speaker A:How do these agents impact that?
Speaker A:Because I, if I could have a hundred of those versus the 15 I can fit around my desk like it seems like I could all of a sudden I could see deeper into the market and I could, I could see it real time because I can't scale my attention beyond what I can look at humans, we can pay attention at A time I can flip between what I'm paying attention to very quickly, but I still have to flip between a move.
Speaker A:How do you see these multiplexing that capability and, and do you think we're going to get to the point where a trader can say I need to see this insight from this data source.
Speaker A:I need to, I need to monitor between these, these situations and when anything like hits that sort of threshold I need you to tell me and either talk to me literally with voice or give me an alert or, or surface it in a way.
Speaker A:I don't know how long it's going to take to where we allow that agent to actually make the trade.
Speaker A:Which obviously the Flash boys, if you read some of those, I mean people have been doing this in exchange traded instruments for a long time based on algorithmic trading with these hedge guys from.
Speaker B:Deep Seeker trying to do.
Speaker B:Right.
Speaker B:Like it's, they're doing trading there.
Speaker B:I mean, yeah, I think but what.
Speaker A:We do is different, right?
Speaker A:It's, we're for physical.
Speaker A:I mean these are most of our, I mean we, we do obviously use derivatives and exchange traded instruments for risk management and hedging.
Speaker A:But the physical work that our industry does to move all this product into the refineries for crude and out as product and down the pipelines and to a custom on a truck and into a C store tank and through the dispenser.
Speaker A:I mean all the trading and insight associated with that is it's a physical thing and it's all bespoke, right.
Speaker A:Like when I do a trade with someone, we don't actually.
Speaker A:It's not an exchange traded instrument.
Speaker A:It's done in ICE chat, right.
Speaker A:It's like it's literally a chat between humans doing the trade.
Speaker A:I just see that as being far different way of bringing AI into the capabilities versus what sort of the high frequency traders have done on the stock exchanges.
Speaker A:Those are all standard instruments that are electronically traded and settled.
Speaker A:Ours are all bespoke text transactions.
Speaker B:Right.
Speaker B:So, so the way I think about this is like.
Speaker B:And you really kind of hit it on now the reason they have so many screens are because they have to pay attention to a lot of stuff, right?
Speaker B:And you can only pay attention to so many things at the same time.
Speaker B:So really they will miss things, right.
Speaker B:That ordinarily maybe they wouldn't have, right.
Speaker B:So wouldn't it be awesome if there's somebody or something, right.
Speaker B:Could watch certain conditions, right.
Speaker B:And then really elevate, you know when something's really important or interesting to the trader.
Speaker B:Right.
Speaker B:But could you do that with traditional software?
Speaker B:Sure.
Speaker B:Like I can write about your code, abc this, do this.
Speaker B:But the problem is again it's very static, right?
Speaker B:It's, I'm writing rule, static rules rather than if I could tell the AI, like look, here's what I'm trying to do, right?
Speaker B:Here's my objective, right?
Speaker B:And these reasoning models can literally now think through problems, right?
Speaker B:And then when you see this data, this data impacts like this, when you see weather in harbor or something, right?
Speaker B:And something is going on, take a look at what we've done with these trades before and just give me an alert, tell me something, I need to look at something, right?
Speaker B:So I think that's kind of the first phase of this where we really just enable this easier insight, you know, more, more ability for trader to do multiple things and watch something 24 7, right.
Speaker B:There have been instances where traders like you get asleep, right?
Speaker B:You go on vacation, right?
Speaker B:And there's, you might get upside down on a trade because of some, some market condition.
Speaker B:You have to go adjust right Now I can at least tell you like look, you're going to have to change this because you're going to be, there's going to be, this is really going to end badly, right?
Speaker B:So, so it can help you I think deal with risk as well, right.
Speaker B:And then yeah, eventually maybe we get to probably we'll get to like sort of more of that automatic trading.
Speaker B:Although right now like my view is just hey, keep the human in the loop, right?
Speaker B:Let the AI do some of the mundane things, some of the basic things, right?
Speaker B:It's already way better than just some Python script you write to look at weather and give you an alert and then, and then, then kind of as you progress, you know.
Speaker B:Yeah, I think those things maybe will make, make some decisions, even trade commodities, you know.
Speaker A:Well, how fast?
Speaker A:I mean with, there's obviously the, the underlying technology is moving incredibly fast.
Speaker A:Our industry was 100 plus years old.
Speaker A:We don't, we don't, we're not known for moving fast and there's good reasons for that.
Speaker A:Like you don't move fast in a refinery because you blow it up or you hurt people and you got to be very, very deliberate.
Speaker A:You have to be very safe.
Speaker A:We're moving.
Speaker A:You know, it amazes me that we can, we can transport via truck, just basically explosive material all over, through rush hour traffic and, and keep people safe.
Speaker A:So it's, there's, there's darn good reasons why we do things in a methodical fashion in our business.
Speaker A:How do you see, how's this different, right?
Speaker A:Is there.
Speaker A:I've been thinking about it from a, just even an enterprise architecture standpoint of, I mean, when I have to go through an ERP transition, it's no one wants to, no one wants to do that, right?
Speaker A:It's multiple year project.
Speaker A:It cost me a fortune.
Speaker A:I don't really have a lot of upside.
Speaker A:I'm more like have to do it right?
Speaker A:Or even when people went through the deployments on CRM or like we've talked about trading a lot with the ETRM or CTRM systems.
Speaker A:Those are very, very expensive.
Speaker A:They take a long time.
Speaker A:We're in an industry already that doesn't move very fast.
Speaker A:I mean, you compare like the point of sale evolution in restaurants with toast and those types of things versus the point of sales that we're limited to in our business because the darn things are tied to the dispensers and they got a fuel controller and we're handling this different kind of material.
Speaker A:We're not just serving up latest special for dinner, right?
Speaker A:Which can be automated and evolve at a very different pace.
Speaker A:Do you think this is different?
Speaker A:Like, because it's not.
Speaker A:Is it replacing systems of record?
Speaker A:Is it, is it, is it a rip and replace?
Speaker A:Is it?
Speaker A:Or is this really something that is more like hiring a new employee that just knows how to use those systems better?
Speaker B:Yeah, so, so I, I think it's not rip and replace, right?
Speaker B:So previously, like let's talk about, so the erp, right?
Speaker B:So it's probably somewhere, you know, things were in like a physical ledger, right?
Speaker B:Then you probably moved it to Excel or had some kind of like an accounting software.
Speaker B:They end up saying, well, we need general ledger.
Speaker B:But all these other parts of our sort of accounting.
Speaker B:So we have an erp.
Speaker B:So those were rip and replace, right?
Speaker B:You took whatever you were using, you threw that away and you installed a new one.
Speaker B:Took five years to kind of get it up and running.
Speaker B:Maybe three of you were lucky, right?
Speaker B:And up you go.
Speaker B:And so I don't think that's necessary, at least at this stage, right?
Speaker B:So because these tools are able to essentially read, understand human language, right?
Speaker B:Automate workflows, they can actually plug into existing systems.
Speaker B:Because I've never met a CFO that says, well, I'd love to, I'm just excited about replacing my ERP system, right?
Speaker B:They're all dreaded because it just, it's a nightmare.
Speaker B:And so don't do that.
Speaker B:Keep your system record where it is, but then leverage the data Integrate the data with some AI layer and then leverage that to do certain things.
Speaker B:Now the interesting thing is, well, once all of that happens, what is the system record?
Speaker B:Is it just like a database with data and the UI goes away?
Speaker B:Because AI is making a lot of these recommendations and decisions where I can just talk to it.
Speaker B:I'm like, hey, make this trade or hey, I need to dispatch a truck.
Speaker B:I got these customer orders, right?
Speaker B:So take a look at this system of record.
Speaker B:Find all the orders in Houston, right?
Speaker B:Find the drivers, look at the compartments of trucks and everything else.
Speaker B:And then just once, you just dispatch it for me and then show me the results, right?
Speaker B:So now all of a sudden, yes, the system of records are there, but they're almost like databases, right?
Speaker B:On top of which we can layer other things and we'll see if this goes that way.
Speaker B:But the point is, don't.
Speaker B:Don't take out this one system and put a new one in, because it has AI, just leverage what you have and then build on top of it, you know?
Speaker A:Yeah, I think that's, I think that's going to make things move faster because you don't have to, you don't have to redo your whole enterprise architecture and all the pain and suffering expense of that.
Speaker B:Honestly, this works.
Speaker B:I've done that in the past where we work with a bunch of like these traditional ERPs.
Speaker B:This was like pretty mobile.
Speaker B:And the customer says, hey, we need a, we need this to be mobile.
Speaker B:Like, okay, well don't, don't replace it.
Speaker B:We'll just add to it.
Speaker B:Right?
Speaker B:So we added a mobile component.
Speaker B:Talk to the APIs or the database directly.
Speaker B:And it's kind of very similar to where you can add to existing products, existing systems, integrate them in a better way.
Speaker B:Right.
Speaker B:With sort of AI being in the middle of it, you know.
Speaker A:Yeah.
Speaker A:I think what's helping me is to think about this more as an hr.
Speaker A:Org design personnel type of mental model, right?
Speaker A:Where it's like, out of.
Speaker A:A friend of mine asked me every day, well, where's.
Speaker A:Where are you going to use this stuff?
Speaker A:And I said, well, pull out your org chart.
Speaker A:Any place you have an org.
Speaker A:Any.
Speaker A:Any box on that org chart that's got some of your best people in it or any of your people in it.
Speaker A:Like you're going to have AI there because essentially they are now your new digital employee that can.
Speaker A:They're not a system.
Speaker A:It's not, it's not a.
Speaker A:It's not just a piece of software, right.
Speaker A:It like has real intelligence.
Speaker A:It can Read, you can teach it context, you can give it guidelines for how it makes decisions.
Speaker A:But I think on again, turning those loose autonomously is, is a ways off, not that far off, but it's certainly not like you said, we got to keep the human in the loop.
Speaker A:But I see it where, whether it's an HR specialist trying to do a comp review or reviewing the job description, I mean it's, I think the league of the legal profession.
Speaker A:So if you've got a law department and you've got a handful of attorney and they're having to employ paralegals and then you've got all your outside counsel that when those bills show up for me, I'm always disappointed at how expensive they are.
Speaker A:I think you can get a lot more done at every layer and every box on that org chart by empowering your employees to just get a heck of a lot more done because they've got a, they're augmenting themselves essentially.
Speaker B:The digital employees are like a really.
Speaker B:Right analogy, right?
Speaker B:And it's funny, I actually heard some VCs talking about looking at sort of org charges like, well, where's your, where's your digital employee?
Speaker B:Where's your AI?
Speaker B:Right?
Speaker B:So the first time I've kind of saw like yeah, that, yeah, that should be on org chart because they're kind of performing a function, right?
Speaker B:And that function is, could be just your subject matter expert to the real human employees or actually you're performing some tasks right now.
Speaker B:The difference is it never sleeps, never, never takes a vacation.
Speaker B:You can, you're only limited by amount of compute.
Speaker B:So you know, you can have one of these ten hundred thousand, doesn't matter, right?
Speaker B:And they just keep cranking through whatever the, they were designed to do.
Speaker B:So that's the, that's the difference.
Speaker B:Now where organizations can scale, like, let's just forget AI, forget technology.
Speaker B:Let's think about it as a, in terms of employees now I can scale my business, I can double it in a year, right?
Speaker B:But then I don't have to double my cost, I don't have to double the headcount, right?
Speaker B:And so all of a sudden we'll see these businesses do amazing things, right?
Speaker B:And drive profitability because they're able to leverage this technology.
Speaker B:Maybe not everywhere, right?
Speaker B:Still human, in a loop, but in a smart way.
Speaker B:And I think that's what, that's what it's really all about, you know?
Speaker A:Well, if you had to guess where we're at in five years, what would you say?
Speaker A:Like five years are we going to see.
Speaker A:I mean, we've worked.
Speaker A:I mean, are we going to go from the home office store support center for a.
Speaker A:For a thousand store chain?
Speaker A:Do they need a thousand people in the home office?
Speaker A:Can you see them getting by with 150, or is it more the other side where they still have a thousand, but now they're running 5,000 stores instead of 750?
Speaker B:Yeah, yeah, yeah.
Speaker B:Well, I think obviously product skating is a hard thing to do, but I think this is one.
Speaker B:This is moving super fast, right?
Speaker B:So I think it'll probably go faster than people think.
Speaker B:At minimum, I think it's literally digital assistant for every employee, right?
Speaker B:You got an employee that have their own digital assistant trained for that job, right.
Speaker B:To help them.
Speaker B:It's going to be any modality meeting, text, voice, video, avatars.
Speaker B:I'm talking to you.
Speaker B:I may not even know that whether you're real Doug or Digital Doug, like the real guy knows also.
Speaker B:It's going to be to that level, I think, so obviously infinitely knowledgeable about certain industry, can learn new things, right?
Speaker B:Completely sort of personalized to me as a human using that.
Speaker B:And I think that's going to be very, very quick.
Speaker B:So in terms of the organizations, back to your question.
Speaker B:Yeah.
Speaker B:I think you'll be able to sell more fuel and dispatch more loads and do more trades, way more than you ever have without increasing headcount, Right.
Speaker B:You'll be able to do things where I've never met a company that said, man, like we have way too many employees, right?
Speaker B:They're pretty much not doing anything other than maybe the government, right?
Speaker B:Everybody's always strapped, right.
Speaker B:Everybody's always busy and running and trying to run the business.
Speaker B:Well, now maybe those employees can actually think strategically, create better relationships with customers because we talk about a lot of AI and technology, but it's going to be a while before AI can actually create deeper relationship with someone.
Speaker B:Right?
Speaker B:And so customers figure out what really moves their needle and then use this technology like you would hammer, right?
Speaker B:It's just another piece of tech that helps you be.
Speaker B:Be more more efficient, you know.
Speaker B:So I think that's where it's going to be everywhere, like every piece of software, every process will be embedded.
Speaker B:Some of it will be where you don't even really know the AI's underneath it.
Speaker B:Right?
Speaker B:Like if you're, if you're today writing email and all of a sudden you have like a nice auto.
Speaker B:Autocomplete, right?
Speaker B:Well, that's AI underneath it.
Speaker B:But the tech kind of goes away sometimes and you're like, oh, great.
Speaker B:Like, I now have a good suggestion.
Speaker B:So I think that's, that's how it's going to be, you know.
Speaker A:Good, good.
Speaker A:Well, I'm excited about the path ahead.
Speaker A:Vlad, thanks for kicking off our Fueling AI podcast series with us.
Speaker A:Please come back.
Speaker A:I hope to have you back over time and we can track on this, track you on this journey and I really appreciate the time and, and for those, our listeners out there, we' we're going to try and keep it real and stick to things we know in our industry.
Speaker A:So there are probably lots of questions about this technology and about this, this trend that we don't address.
Speaker A:But if it's going to impact the energy business, the fuel business, the convenience business, things where we all live and work, we're going to try and bring you that, bring you that news, translate it into language that makes sense for us that live in this business and help you find a way to add value to your enterprise with this new tech.
Speaker A:So appreciate you joining us today.
Speaker A:Thanks for being here, Vlad.
Speaker A:Take care.
Speaker B:Take care.
Speaker B:Can't wait to hear more about the other leaders that I can learn from them and looking forward to kind of hearing all about these different new episodes coming up.
Speaker A:Great.
Speaker A:Thanks, Vlad.
Speaker B:Take care.
Speaker A:Take care.
Speaker A:See you guys.
Speaker C:The AI revolution is here.
Speaker C:Are you leading the charge or falling behind?
Speaker C:Fueling AI with industry CEO Doug Hawk is your blueprint for success.
Speaker C:Subscribe now and supercharge your fuel and C store business with the power of AI.
