Welcome to another episode of Growth Challenges for Manufacturers and How to Overcome Them. I’m your host, Katherine Seymour. This episode is brought to you by MacRAE’S trusted by North American businesses for over 100 years. As a leader in B2B digital marketing, we help industrial and manufacturing companies struggling with stagnant growth or lead generation. By leveraging advanced AI and automation,
In our SEO and lead generation programs, we help you appear more prominently in search and AI results like chat GPT. This drives significantly more qualified traffic to your website, resulting in stronger lead flow and increased revenue, crucial in today’s highly competitive digital landscape. Learn more at macraes.com today. Today I’m joined by Harsha Pei, co-founder of Athion Solutions, a company that provides cutting edge technology consulting
and services to help businesses leverage AI, data and digital transformation to drive growth. Atheon Solutions partners with organizations to build smarter, more scalable operations from advanced analytics and cloud strategies to custom software deployment.
With a strong focus on innovation, Harsha and his team are helping businesses adapt to rapidly changing markets to unlock new opportunities through technology. Harsha, it’s great to have you on the show. Welcome.
Thank you for having me on your show.
Thank you so much. Harsha, let’s get started with your background. What led you to co-found Athion Solutions and what excites you most about working with them today?
Right, so let me give you a little bit of my background first. I have spent about three decades at the intersection of business operations and traditional technology, right? So in 30 years, I think I’ve been very, very fortunate to have worked with large Fortune 2000 companies. ⁓ I have built businesses, run businesses, transformed businesses out of roughly 26 different countries in those three decades. And that made for a very, very exciting career, right? ⁓
Yeah.
Yeah, but be honest, I think it’s the last two years that have been the most exciting. And you probably know exactly what I’m going to say because it’s over these two years that AI really came of age, didn’t it? And I think what AI has really done was really to open new vistas.
Yeah.
to accelerate innovation and transformation. To be honest, that’s really the calling that I had. So me and my three other co-founders decided that this was the right time ⁓ and the right place for us to build out Athom.
Mm-hmm.
Amazing, amazing. From your perspective, what makes AI and digital transformation space so dynamic right now?
Yeah, I would say follow the money or to be more precise, follow the smart money. If you look at the statistics, I mean, go back to the year 2023. And if I have my numbers correct, those roughly about, I would say, give or take about 300 billion dollars that went into funding of, you know, new businesses, right? So your venture capital, hedge funds, that was a total of about 300 billion.
Right.
And within that, I would say about 25%. So that’s about 75 billion went into new age technologies such as AI. Okay, so let that be our baseline. Getting into 2024, that number went from, so the total funding again stayed flat at roughly about 300 billion. And the 75 billion that went to AI suddenly started ballooning up. So that number became about 30 % or so. So roughly about a billion, about $100 billion, I’m sorry.
Yeah.
And now we are talking to each other in September of 2025. So nine months in the year, the number that started at 25%, which became 30 % is now, believe it or not, 50%. So.
Right.
That tells you something, that smart money is looking for smart tools, smart capabilities, smart solutions that could drive innovation, ⁓ make change real for us. So I would say, think that was probably the first indication for us that there is a lot of dynamism that this new age technology represents. think of the big companies out there, the Open AIs, ⁓ the Grocks of the world, the Anthropics of the world.
Hmm.
Yeah.
And we’re seeing new models coming up literally every single week, right? They’re ⁓ much more larger, much more faster, a lot more economical, and therefore a lot more easier for people to actually go out and adopt. And I think that’s what’s actually enabling innovation and the ability to build a lot more faster. So I’d say that’s just to give you a sense of how dynamic this space is for us today.
Right.
For those who might not be familiar, can you describe what your company does and what industries you serve?
Sure. So our company tagline is reimagining operations.
And what that really means is, you know, we’re looking to bring, you know, some of those new age technologies that I spoke to earlier, right? So it’s artificial intelligence, machine learning, robotics, process automation, to actually go out and completely transform how a wide variety of business operations today ⁓ actually operate. To think about it, you know, there’s so many of these common functions across, you know, sales, marketing, accounting and finance, ⁓ procurement, compliance, and these…
Yeah.
cut across industries beyond just manufacturing alone. And I think that was the real opportunity for us. As a company, we four co-founders, as I mentioned, two of us based out of New York, myself and another co-founder who are based over here in India. And all four of us come with a minimum of 25 years of…
experience in that sense we’re practitioners. So our proposition to our clients really is, you we’ve kind of been on the other side of the table. We know what it means to be able to deal with legacy, to be able to deal with transformation, and most importantly to deliver to some of those outcomes that, you know, we’re all expected to deliver. So that said, I think the first industry that we really picked up was financial services. ⁓
Ray.
Yeah.
huge industry, ⁓ highly regulated, ⁓ high value and high volume. So I think that really allowed us to build capabilities that were put to the test.
Having done that very, very successfully over the last, I would say, year or so, ⁓ we were then encouraged to take this beyond just the world of financial services. So whether it’s going into a manufacturing company and helping ⁓ finance and procurement functions automate, or going into technology companies and looking at how ⁓ some of their sales and marketing processes work and leverage AI to be able to accelerate pipelines for them, I think we’re now cutting across industries ⁓
Mm-hmm.
delivering value across a number of different functional areas. So that’s what we do here at Aetha.
Very cool. Businesses everywhere are trying to adapt to technological change. What do you see some of the biggest challenges companies are facing and how is your technologies and software is helping them?
really say there are two fundamental types of challenges that companies are dealing with today with respect to technology change. ⁓ The first is really a bunch of technical challenges in themselves, right? So like I said,
over the many years and decades that people have built on the back of traditional technology, they inevitably created what I would call technology debt. So there’s a lot of legacy that’s sitting there, that’s data sitting in silos, that are systems that don’t really speak to each other. ⁓ And I think there’s a patchwork of some of these, which I would just refer to as technical issues. And ⁓ mean, honestly, I don’t worry so much about
Yeah.
Mm-hmm.
technical issues because I think there are solutions for those. The ones that I worry more about are what I would call cultural challenges, right? things such as resistance to change. We’ve always done it. This is what we’ve always known. This is what’s worked for us and all that is now changing. So I’d say it’s the leadership challenge.
Yeah!
⁓ It’s the challenge around organizational design and re-skilling of people to be able to deal with some of these challenges that I personally think are the slightly more difficult to resolve.
Hmm.
Right,
right. Many organizations worry about balancing innovation and cost and efficiency. How does your solutions help them overcome that challenge?
Well, one of the things that we do when we engage with any prospective client is really drive this whole process of engagement through a very consultative and collaborative framework. What that really means is as practitioners,
we come in with a bunch of things. Number one, with a lot of domain expertise within that area. While earlier I did mention that AI does have universal applicability, cuts across industries and functional areas. But when you start to really peel the layers of the onion, you soon realize that there are a lot of nuances that exist.
Yeah.
in your specific business, with respect to your specific customers, how your processes are laid out, your technology and your workflows. So therefore, having that highly collaborative approach where we come in, do a very quick discovery to understand all of those nuances.
Mm-hmm.
and then baseline what your current performance is today. So how are you really performing ⁓ financially? How are you performing from a quality standpoint, from a customer experience lens as well? And then based on that, build out almost a custom transformation roadmap for our clients. So to come out and say that if this is where you are today, this is the art of the possible, these are some of those technology interventions that we could bring and solve for each of those specific problems.
Ray.
quantify the benefits and of course the cost of that transformation and therefore in that process literally show them and in some cases even underwrite what the ROI for that technology change could look like. So I really think it’s very very unique in how we consult, how we baseline and how we go out and guarantee ROI for our clients today.
Yeah.
Amazing. With the pace of technology advancements, ⁓ how do you work around the demand for AI-driven solutions compared to traditional IT solutions?
There’s first of all a huge demand out there, right? And the moment you use the word AI, it’s only natural like you did to ⁓ have this quick smile, but also a bit of a roll of the ice. So I will not deny that there is some skepticism out there as well.
Yeah.
Yeah!
We all probably saw the recent MIT study which seemed to indicate that over 90 percent of AI initiatives taken up were either shut down or they’re to fail. Having learned from what works well and more importantly, what does not work well, here are a few of the things that I would go out and tell anyone who’s interested in going through this journey. Number one,
Mm-hmm.
Yeah.
What I have found is that with AI, the ability to go from, let’s say, zero to 60 is very, very quick. So the ability to prototype solutions, the ability to very quickly test, validate, and confirm if this is something that could work has really been accelerated. So I think you could do that in mere weeks, if not days.
Right.
Going from 60 to I would say 80 to 90 % is where I think you start to encounter a number of edge cases. It’s like I said earlier, it’s what is it that makes your business unique, your specific proposition to your customers, your technologies, tools, your workflows. And that’s where you start to realize that prototyping took you to a certain point, but there’s a lot more in terms of being able to test engineer.
Mm-hmm.
Yeah.
⁓ redesign and reinvent, right? So I would say that’s the first big lesson. The second, I think, key lesson that we’ve learned and we always share with our clients is pick your use cases, right? There’s almost this perception that AI could do everything today. It’s like this little magic. ⁓ Reality is far from it. AI by nature has proven that it is a probabilistic technology. Unlike, you know, good old tech that we’ve all come to
Raise.
see, understand, and use, which is really very deterministic. ⁓ Now, things have suddenly changed. You’ve got issues around hallucination, you’ve got issues around data drift. ⁓ Therefore, AI, at least at this stage of its ⁓ maturity, I would say it doesn’t necessarily apply itself to every single kind of use case. So be very selective and careful about where and how you go out and implement AI as a potential solution.
Mm-hmm.
Thirdly, think the most important thing is with all the hype that’s surrounding AI right now, there’s probably this belief that AI could do 100 % of everything that humans do. And a lot of failures that we’ve seen is where people almost prematurely took humans out of the loop. ⁓ We’re convinced and strongly believe that while there are a bunch of things that AI can do really well,
Right.
Mm-hmm.
there will always be the need for human ingenuity, ⁓ the need perhaps for that warm handshake in areas such as sales and marketing. And beyond simple quality checks and affirmations, there’s probably the need for a sanity check, which only humans with the right experience and the depth of domain could go out and bring. So while everybody today talks about large language models, ⁓
Yeah!
Yep.
That’s great. We’re talking about large logic models where you certainly do need human expertise to kind of rein in what AI can and cannot.
For sure. And it’s a combination of AI being a tool that helps experts be better at their jobs, which is amazing.
Absolutely.
You offer a wide range of services. Can you walk us through how you help businesses identify the right solution for their needs? I know you mentioned that you kind of go through a really long consultation phase. Is there a specific site you touch point on those?
Yeah, contrary to popular belief, we try and kind of make it as efficient as we can. So while I did mention that it is a domain led ⁓ industry deep consultative engagement, but just by virtue of the fact that we bring in people who got decades of experience having been there and done that before actually helps us accelerate the process. So ask the right questions, ⁓ bring the right checklists, ⁓ bring the right artifacts for the right problem statements. So I think the
Yeah.
Yeah.
domain expertise that’s built into that consultative and collaborative engagement model actually helps us accelerate the entire engagement by itself. other thing that ⁓ I will say is, ⁓ today we see a lot of technology companies out there and what they’re trying to do is just sell you a tool, right? So it’s 14.99 ⁓ subscription per month and then ⁓ they disappear immediately.
Mm-hmm.
for sure.
Yeah. Yeah.
⁓ The other thing that we’ve realized with AI is it’s not just a sale. It’s a journey of engagement and therefore what we’ve almost created is like this professional services layer that almost invisibly sits behind our platform where we’re constantly watching how our customers are ⁓ utilizing our tools ⁓ and how are they actually delivering or leveraging and getting benefits out of our services.
Mm-hmm.
More often than not, we’re actually going back to them and saying, hey, we looked at how you did something and here’s a quick unlock, or here’s another quick hack, or here’s a best practice. And in some cases, like I said, there’s also this belief that AI should be doing everything, which clearly at this point, we aren’t there yet. ⁓ To be able to then go out and also kind of handhold our clients around areas that you necessarily cannot have AI do everything for you. So to be honest, we’re…
Yeah.
Mm-hmm.
To us, AI isn’t just about artificial intelligence. For us, AI stands for applied intelligence. It’s about where you can see and extract the best benefit of what AI can deliver for you today.
⁓ Cool. What steps do you take to ensure that your solutions are delivered and scalable and secure for your clients?
So I’ll try and answer that in as non-technical way as I can. So for most of your audience, people who would appreciate business speak. ⁓ One, given the fact that we lead with industry domain knowledge and expertise, we spent a lot of time and effort in building what I would call operating artifacts, right? So while AI in itself, large language models tend to be a bit of a black box, ⁓ the ability to create industry specific
Mm-hmm.
Right.
whether it’s business rules, ⁓ masters, ⁓ validations, checks and controls, ⁓ are all built into these little artifacts that we think are our differentiators. ⁓ So you’ll actually see us come in with, let’s say, a golden data set, or like I said, business rules that enable you to be able to accelerate and create for a seamless workflow of processes. So I would say that’s one thing. ⁓
Right.
You also spoke about things like scalability, ⁓ security, and sustainability. The one thing that I almost take pride on is having learned from some of the things that did not go well for us in our early days, ⁓ the word that I use for it today is design for failure. So anticipate what can go wrong. Think of…
Yeah
capacity requirements and constraints and therefore, ⁓ add resilience and redundancy upfront. ⁓ And in that sense, you’ve almost preempted ⁓ even before something could potentially go wrong. So these are some of the things that I think we do quite well to ensure that our solution is scalable.
Mm-hmm.
Yeah.
If you could instantly solve one common obstacle for companies when they face their digital transformation, what would it be and why?
have to say resistance to change. ⁓ We kind of briefly touched upon this. There are so many organizations and some of those industries which for a variety of reasons were a little slower to kind of embrace transformation where for as long as you can imagine there was just this one way that we’ve always been doing it. ⁓ And it’s worked till the point that it did not work. And as businesses scale, and I can…
Yeah…
Hey.
tell you so many manufacturing companies, ⁓ so many retail companies ⁓ who kind of stuck to that age old way of how it’s always been done have now struggled to compete with new age, born digital, AI native kind of organizations. ⁓ The other thing that you’ll notice when it comes to transformation is transformation typically starts with this really smart group of really smart, capable people.
Yeah.
had almost sit in a corner office or in an ivory tower. The reality is, transformation can start there, but very quickly needs to be embraced at a grassroots level. So to go back to your question, if there was one common obstacle that precedes really this, it’s the resistance to change. It’s about transformation that’s sitting in ivory towers, that’s sitting in small little ⁓ pockets that we need to break.
Yeah.
Mm-hmm.
⁓ and that bring into the grassroots of mainstream organization itself.
Yeah.
sure. Looking ahead, what trends in AI cloud or digital transformation do you think is the most ⁓ like the most shape for ethanol to grow?
I would say there’s two or three. So starting just with the foundational infrastructure itself, as I mentioned a couple of times already, ⁓ you’re actually starting to see AI models that are ⁓ much more powerful, ⁓ much more faster, ⁓ much more cheaper or economical, much more smarter. know, better and better reasoning models coming up every day. So I’d say that’s clearly one big theme.
Mm-hmm.
Yeah.
that will continue to kind of watch out for, leverage, bring to our clients. The other one that I think is shaping our growth is this whole notion of what does large in a large language model really mean, right? The first couple of years, people were looking to go from 100 million parameters to a billion parameters and 2 billion and then 100 billion parameters.
rates.
And there was almost a bit of an arms race in terms of who’s got the bigger, more powerful model out there. And to some point where you realize that as opposed to a jackhammer, what you probably need is a surgeon’s knife to solve for that specific problem. So the second big theme or trend that we think will shape our strategy going forward, and for much of the industry as well, is large language models.
Yeah.
Mm-hmm.
versus small language models. Today we are seeing small language models which are probably smarter, more appropriate, more fit for purpose ⁓ that address very specific problems that companies have. So that’s one to watch out for. And the third, and something that’s I think very, very topical and critical for us here at Athon is what I would call vertical AI, right? So as opposed to AI for everybody.
Hmm.
all the time and for every single use case, which I think is a bit of a misnomer, ⁓ we just think the ability to go niche and vertically deep allows us to be able to take specific use cases and solve them very, very effectively end to end. So for example, in the manufacturing space, ⁓ think of let’s say a procurement problem and within procurement, think of a very specific category management or a sourcing kind of a problem.
Mm-hmm.
You don’t necessarily need this massive monolithic gargantuan model to solve for that very specific model. Problem statement, what you need is this laser sharp, highly focused domain deep solution to solve for it. So I would say watch out for vertical AI.
Yeah.
Sounds great. Can you share a recent success story? Maybe a client who you helped solve a major challenge or are skilled in your operations?
Sure, yeah. So as I mentioned, a lot of what we did in last year was really focused around financial services. But as we solved for that problem, we soon realized that at one level of abstraction, regardless of industry, we see the same problem statements across the board. So maybe let me start with what those problem statements were. When you think of classic business operations, right?
The one thing that you’ll always see within operations is, let’s call it unstructured documents, right? So there’s data that’s sitting in an email somewhere. There’s an attachment that’s got the answers to exactly what you’re looking for. There’s a purchase order or there’s an invoice that was sitting as a PDF that was attached to an email or stored somewhere. So regardless of industry, regardless of functional areas, the one…
right.
big issue that we saw, of course, in this particular client’s case, but across the board was the amount of PDFs and Excel sheets that were floating around the system, ⁓ unattended, disorganized, but which had really critical information that you ideally needed to have in your system. So the very first thing that we did with AI was the ability for generative AI today to be able to ingest semi-structured and unstructured file formats or documents.
Mm-hmm.
there were roughly about 80,000 such documents that they would receive on a monthly basis, which would have taken them a small little army to actually process. With the power of AI today, you could do that literally overnight. So all of a sudden now, we had this very elegant, very powerful, highly scalable, very secure model that could do what that mini army did with respect to just dealing with unstructured documents.
Absolutely. Yeah.
At the end of that process, what we were able to get out was clean, organized, structured data. And when we gave that data back to the client, guess what? Like any other client across industries, that data once again went into this highly unstructured tool. Exactly, words out of my mouth. And again, that’s another classic symptoms that we see all the time, which is ⁓ data is unstructured, data is siloed, data doesn’t need.
Cool.
doesn’t always speak to each other. And at the end of the day, you don’t really have business ready data. So decisions that need to be taken, strategies that need to be evaluated are literally waiting on clean organized data that’s kind of distributed across different systems. So the second problem statement that we were looking to solve for was almost create this, think of it as a multi-dimensional analytical platform that virtually connects into
Mm-hmm.
your existing highly unstructured data platforms and provides you with this, know, think of it as almost a cube where you could get different views into your data that answer all these multi-dimensional questions. you know, what, for example, I’m just making this up, but, you know, a management question could be, you know, what’s the profitability on a specific category of merchandise across certain retail outlets?
at a certain period of time. Now, each of these different pieces of data probably sit across different systems, but the ability now to be able to virtually scan through all of that data and be able to produce back for you highly usable, highly relevant business ready data is the second problem that we were able to solve for AI. And when we kind of solve these two big problems, right? So you had this unstructured and semi-structured documents on one hand, and you had unstructured data on the other hand.
Ahem.
Once we were able to kind of meaningfully address that, what we soon realized is downstream processes that were kind of broken up, we were able to drive automation levels significantly higher, therefore saving costs, saving time, and more than anything else, I think just improving the quality quotient of what our clients were delivering.
Right.
Amazing. Finally, what advice would you give to business leaders who are maybe hesitant to take the leap into AI and advanced digital strategies?
Yeah, so a couple of things that come to my mind. ⁓ Number one is, know, implementing changes is difficult, right? Now, whether it’s bringing in AI or bringing in any form of new age technology, you know, we’ve always seen that resistance to change. So the one thing that I would tell people right up front, even before they venture down that path is to get buy-in. Okay, like I mentioned earlier,
Yeah.
Hmm.
There are a bunch of very smart people that typically sit at the heart of driving transformation within organizations, but it’s a really small group that’s unfortunately sometimes sitting in those ivory towers or those corner offices. think buy-in is about engaging with all of your internal stakeholders, ⁓ not just your senior decision makers, but the actual user groups by themselves. ⁓ communicate openly. ⁓
Right.
evangelize the idea, get that buy-in across the broad cross-section of an organization where you’re looking to drive some of these changes. That’s number one. ⁓ Number two, I also talk about strategic clarity. AI is probably at the peak of its hype cycle at the moment, and I cannot tell you how many cases we’ve had of someone who just says, we need AI in, and we need to implement AI, or AI is a management mandate today. ⁓ Very rarely have I…
got a clear answer for why is AI a mandate or why is AI critical at this point of time for your business? And I think being able to answer that why provides for that strategic clarity. And once you have that strategic clarity, I think a lot of other questions also automatically get answered, which is, you know, where does it make sense? Where does it not make sense? And this goes back to the whole idea of applied intelligence as opposed to just, you know, general artificial intelligence.
Right.
⁓ The last thing of course that I will also talk about is understand the risks. While of course AI does represent tremendous potential universal applicability and it’s growing really really fast. So the next time we talk about AI, whether it’s the next week, the next month or the next year, something tells me there would be significant ⁓ improvement from where we are today. But that said, there are some very real risks in so far as
Right.
Hmm.
⁓ how these models work. So there’s an angle of ethics, there’s an angle of explainability, there’s an angle of hallucination and accuracy of what these models will predict. So I would say go in with your eyes open. I guess in closing I’ll say the analogy that I always like to give clients is think of AI implementation almost as if you were going in for a home renovation, a massive.
Right.
home improvement project while you’re still living in the house, right? How would you approach it? So right up front, what you do is make sure that you have a clear blueprint. Even before that, maybe that you’re talking to experts. Get in the experts, people who understand this better than you, people who’ve done this, you know, learned what works, what doesn’t work, have the right plan, right? So create that blueprint. And because you’re already living in that house while you’re looking to make changes, ⁓
Bye.
sequence it logically. So for instance, do you start with the kitchen? Do you start with the lounge in the living area? Do you start with the bathrooms? You don’t do all of it together while you’re living in the same house, right? There is no way. And have a plan B. Always have a plan B. So, you know, while something’s working at some point, you know, there will be delays, know, things will break, things will fall. ⁓ How do you ensure that you always have that other room or you have a backup plan if something doesn’t go to plan?
Wait, there’s no way. There’s no way. Yeah.
So I would just say, do something that you’ve always done in your normal life and I you’ll be fine.
Amazing. Well, thank you so much for joining us today. You gave us amazing insights, and we truly appreciate you joining the podcast. And for everyone listening, thank you for watching. And we’ll see you on the next episode of Growth Challenges for Manufacturers.
Thanks once again.
Contact us to explore how we can help your business grow.
