Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Beyond Chatbots: How AI Agents Are Transforming Business | Rajesh Sinha (Fulcrum Digital)
116 Plays8 days ago
How do you build a $100 million enterprise tech company - without raising a single dollar from venture capital?
In this episode, host Akshay Datt speaks to Rajesh Sinha, Founder of Fulcrum Digital and Culinary Digital, shares his remarkable story of scaling two global businesses to ~$100M ARR - fully bootstrapped.
From building a customer-funded SaaS + services hybrid model to leading the shift toward autonomous AI agents, Rajesh reveals his frameworks for disciplined growth, tech innovation, and global expansion.
He also unpacks how AI agents - powered by SLMs (Small Language Models), governed by secure enterprise frameworks, and orchestrated via Fulcrum's proprietary FT Rise platform - are set to reshape the future of business software.
Whether you're a startup founder, enterprise AI strategist, or services operator, this conversation is packed with insight on how to scale without capital, differentiate in crowded markets, and win in the age of AI.
Key Topics Covered:
👉How to bootstrap a $100M SaaS and digital services company
👉The evolution of Fulcrum through 5 tech eras: Web 1.0 → AI economy
👉Why vertical and horizontal AI agents are the future of work
👉How Fulcrum is using SLMs vs LLMs for cost-effective AI delivery
👉The mechanics of AI agent governance, pricing, and orchestration
👉Fulcrum’s global delivery model: India, Brazil, Argentina, and beyond
👉Navigating enterprise sales, customer concentration risk, and culture
👉Why Rajesh now sees the value of investors - but didn’t need them
#BootstrappedStartup #AIAgents #DigitalTransformation #TechEntrepreneur #StartupGrowth #CustomerFunding #IntelligenceAmplification #PlatformBusiness #EnterpriseAI #FounderThesis #TechServices #ScalingWithoutVC #AIStrategy #BusinessAutomation #TechLeadership #StartupPodcast #Entrepreneurship #SaaS #GlobalBusiness #techinnovation
Disclaimer: The views expressed are those of the speaker, not necessarily the channel
Recommended
Transcript
Understanding Client Needs vs Wants
00:00:00
Speaker
It's always a gap between what do you need versus what do you want in every business. And if you have an ability to help them clarify their need, I think you are one step ahead in the sales process.
Embracing Discomfort for Growth
00:00:11
Speaker
So breaking away from that your own self nature that let's not be comfortable, be uncomfortable is the recipe for growth. Nice guys don't scale their businesses.
00:00:23
Speaker
ah you You need to be willing to be ruthless as a founder to really... get to scale. and Do you agree with that? We could have been a billion dollar company by now.
00:00:34
Speaker
We could have been a half billion dollar company by now.
Introduction to Rajesh and Fulcrum Digital
00:00:52
Speaker
Rajesh, welcome to the Founder Thesis Podcast. ah You have founded and run Fulcrum Digital in addition to a couple of other subsidiaries.
00:01:04
Speaker
And this is an entirely bootstrapped venture. Phenomenal story. ah Can you start with helping me understand the scale and scope of all the business ventures that you're running today?
00:01:18
Speaker
Yeah, Akshay, thank you for having me at your show. um it's ah It has been a great journey. I call it a marathon journey. We built two companies, Fulcrum Digital, which just stands as a global company with 1,500 employees, having a five business parts with...
00:01:36
Speaker
large logos with 20, 30 billion dollar customers in our portfolio. And all these employees are based in the US, UK, Europe, Latin in America.
00:01:49
Speaker
and India. So building a global delivery model while creating a value for customers locally in the market has been a phenomenal journey
Bootstrapping and Revenue Strategies
00:02:00
Speaker
for us. And the second company we started Culinary Digital by which we are touching about 3 million lives of patients and students across the globe.
00:02:13
Speaker
using our SaaS software in commercial kitchen management. So we'll talk a little bit about it. But these are the two companies. We have built a global company. we are headquartered in the US. What's the combined ARR for both the businesses? um You know, we are... The way we are measuring is number of lives we are transforming right now. So we are transforming 3 million lives. So, um you know, that...
00:02:42
Speaker
Real ARR, we are not offering it on the show. But just broad, like say, are you in the $10 to $50 million dollar range or $50 to $100 million dollar range? Very, very broad. Yeah. So we we we are in the trending between $10 to $20 million dollars ARR business.
00:03:04
Speaker
and Combined for the two. Okay. Yeah. No, this is just a culinary digital. Fulcrum Digital is, you know, approaching to be 80 to 100 million dollar company.
00:03:19
Speaker
Wow. Amazing. So, and this is all bootstrapped. You have... not raised any VC funding. So you you are today at roughly a hundred million dollar ARR combined ah kind of a business with no VC money, no external funding.
00:03:38
Speaker
Correct. Yeah. Amazing. If you want to put together the business model. okay so so you know that's what i want to learn from you on how to hit this kind of scale with no external funding ah so could you take me through your journey i i believe this is version five uh so you know just take me through uh what was version one version two version three and ah along the way if there are lessons for other builders based on your own journeys Yeah, sure. um
00:04:10
Speaker
When we started, those were the eras of Y2K, where the people believed in making it quick money, getting into Y2K, and we believed in creating a next generation value-oriented company. so but who Who were you at that time? like Were you a software developer? Were you already an entrepreneur? Were you a management consultant? Yeah.
Rajesh's Journey and Business Beginnings
00:04:36
Speaker
Yeah. So I was I worked in Mumbai with as a software sales consultant. So I was a software salesperson selling Informix. ah Later on, we were selling Computer Associates product.
00:04:49
Speaker
Then I got into selling software outsourcing projects, you know, worked with companies like L&T, um Larsen and Dubrow, then came to the US, worked for a small farm, make sure um I was their salesperson.
00:05:04
Speaker
And that's when I got the business idea of the client connects. And and i I felt confident at that time that I'm ready to kickstart the business. Once the inner voice comes in and conviction is there, you believe in it and you jump in like any other entrepreneur, right? So i have a strong conviction.
00:05:26
Speaker
um I've seen the India side, I have seen the US side, seen the variety of customers in the US and that gave us ah clarity and confidence that there is enough market for value creation.
00:05:39
Speaker
And how much capital did you start with? You probably used your savings to start, Rosh. Yeah, it was few few thousand dollars. We had a savings. So, you know, we put those savings in the beginning. You need to have office space.
00:05:54
Speaker
You need to create your own finance department, HR department, few employees on your payroll. um And then the first year we had to recruit about 11 people.
00:06:05
Speaker
You have to first find a first project to fund it when you are bootstrapping it. your funding sources are your customers. So you must have an ability to get one or two customers.
00:06:18
Speaker
Many entrepreneurs think that getting just one customer is like building a business. To me, that was not the recipe. It was like, do you have a recipe of building multiple customers, you know, one after another? As long as engine is not there, we will not have a funding, right?
00:06:35
Speaker
So that was the recipe for our bootstrapping funding approach. So... What was your first deal? Do you remember? first Yeah, it was a large insurance company. They wanted 11 people team to build a um you know one of the portal technology because those days we were into Web 1.0 and we needed to build a Web 1.0 applications on portal.
00:06:57
Speaker
So we had to staff 11 people to build that. And recruiting those 11 people requires funding because by the time you get the customer's money, you have 90 days of cash flow planning to do, right? So you have to use your funding, plan your cash flow, minimize your expenses.
00:07:16
Speaker
That's the... bootstrapping approach, right? And as you are on the journey of delivering one customer, you have a pipelining approach, right? You have to throw seeds in building another 20 prospects, qualify them.
00:07:31
Speaker
And as this project is maturing, you have two more new clients rolling out, you know. So that's the cycle we got into and we became successful, right? And the culture of continuously building marketing, sales, nurturing prospects, funneling, those were my DNA anyways. you know So that helped me and not lose my sight on how to build a sales engine. And that was the funding source.
00:07:58
Speaker
right And then comes the execution. you know That's when you rely on hiring right people with the right skills. I'm not good into managing the project, delivering it. So you need to recruit ah um the right people to complement your skills, right?
00:08:16
Speaker
um In the end, when you start as an entrepreneur, you are the CEO and you are the janitor. You do everything. But as you scale your business, you've got to bring what you are not good at and what your company structure needs.
00:08:31
Speaker
So that's when I got my project manager. ah had a finance person, HR person supporting me. and then the business starts to shape up like like your delivery was happening out of india uh in the beginning it was all in the us so we started in the us and the first team structure we created in the us and then later on when we started to build multiple clients then those customers started asking more team members from india and also one of the large very large insurance company project we were doing and they wanted to start a center in india so one of our employee was leading that initiative so because of that we came down to india and they were willing to start their center in india and that helped us kick start our base in india to expand our presence you know
00:09:26
Speaker
Okay, okay, got Did you stay ah focused on
Market Focus and Customer Strategy
00:09:31
Speaker
one sector? Like your first client was insurance and you ah probably had multiple insurance clients. Has that been the focus area or like, you know, from in terms of how you position the business, do you position it as a specialist in one sector or what is it positioned as?
00:09:49
Speaker
If I have to go and start now, i would do that. But that time I didn't do it. and i was a young guy, I was 28 years old. Whosoever is willing to give me the money, size didn't matter.
00:10:01
Speaker
um Any customer I will welcome, right? Even small customers, large customers, wherever my contacts were, I'm going all over, right? Certainly the interest was large company. In fact, one of the big bank, when they were negotiating,
00:10:17
Speaker
their contracts department with me. And they asked me what's the um proposal value and cost structure. And I had certain dollar value, very conservative figure in my mind.
00:10:29
Speaker
And I didn't get that number. I just asked that person, you tell me and I'll make the business work for you. And she gave me the number, which was double than what I would have even imagined to ask. So sometimes your funding finding happens and I got lucked out. And just i imagine like six, seven people team, double the money while I was even thinking a young guy. And I was almost at the verge of saying this number and just I hold it. And I said, don't worry, I'll work with your budget. And she turned out to be a very big bank.
00:11:00
Speaker
So sometimes big bank. And then I worked on the other side, a small travel portal company, who was very, very tight in paying the money. So even you ask some money, they will not offer you. So I had to deal with both sizes of the company, the company who can pay me double the money, mike more than my expectation, and the company who is not willing to even pay what is needed for the project. So you have to deal with both and you can't say no at this stage. But today we say no to many businesses.
00:11:27
Speaker
We have a very focused approach, and but that was not the case when we started. it oh So for someone who's starting out today, what would you advise him in terms of should you be picky and choose clients and how do you judge which client to choose?
00:11:42
Speaker
Yes, if have to go and start on my own today or any entrepreneur, I will say that. I have made many mistakes and one of the mistakes um is, you know, I try to do too many things, try to scale all sides of the cloud clients and that's not the recipe to grow big faster.
00:12:02
Speaker
Right. It takes a slower pace. So if you stay focused, you define your customer base and stay in your lane, you will grow bigger and smarter and stronger. and that's that's the correct way to grow if you are into manufacturing or finance or insurance or you have certain domain of products stay in that product the market has a lot of noise and adjacency excitement and the good entrepreneurs who do not get excited stay on their course are are having the recipe of hot hockey stick growth
00:12:38
Speaker
Okay. ah How do you judge ah who's a good customer? Because, you know at times as entrepreneurs, you actually need to fire your customers.
00:12:49
Speaker
So, you know, how how do you judge who's a good customer, who's a customer you should look at firing? Yeah. So before you go on to judging customer or judging yourself, the first you need to judge the situation and opportunity.
00:13:04
Speaker
Opportunity is the biggest customer, not the customer is giving you the opportunity. So once you qualify the opportunity, you need to know whether you are capable to deliver on that promise and your promise matches with that opportunity. So that is the winning formula.
00:13:20
Speaker
Then the second layer comes in, who is the owner of that opportunity? Then comes the customer. ah But once you have handle on who the opportunity is, you need to also understand the financial backing of that opportunity.
00:13:35
Speaker
If the financial sources are more secure, more robust, then you are comfortable to do business. Then who is on the seat? is the third question right um because the person who is on the seat might change company may not change right so um so then you evaluate that and the only way you evaluate is by asking the right questions getting the right answers going deeper and deeper don't be shy in asking questions because more you ask better you are you know you're clear about the business problem
00:14:09
Speaker
and then What are some overlooked questions? like Like questions someone who's starting off may not think of asking. Non-intuitive. Yeah. Question is about the common vision of the project because everybody wants to be an order taker. They tell you this is what I want.
00:14:28
Speaker
But the first question is, what do you need? It's always a gap between what do you need versus what do you want in every business. Every customer who is buying is trying to define what he wants.
00:14:43
Speaker
I want this, I want that, want this way. he has a big dead demand. But if you go into the core opportunity and issue what he's facing, a smart service provider or smart vendor who has a product will very clearly ask right question.
00:14:59
Speaker
Actually, your need is something else, but you're asking me 20 different things. Why you want to spend so much of money asking 10 things where your problem is this? And if you have an ability to help them clarify their need, I think you are one step ahead in the sales process and you can close the deal faster. And that's when successful entrepreneurs scale in the business.
00:15:23
Speaker
and Fascinating. Like ah a need analysis is actually the starting point of sales in a way. like i that's how I would say that. yeah In the beginning, I didn't know. But after 30 years of running two great companies, global companies with 1,500 employees, I think I have made enough mistakes to share these expertise now with the audience. Right.
00:15:44
Speaker
So, ah how did it go for the first couple of years? By when did you reach like a ah respectable, let's say, a million dollar plus kind of a revenue number?
00:15:56
Speaker
It took us a year and a half to two years when we crossed about a million dollar revenue. um And I think we had office in Fanwood, a small town in New Jersey with five room suite with an open area where developers were sitting in. And when I saw that, I have all the right people on the seats i had a hr person had a dedicated finance person a delivery person i could see that now company has a clear vision have a right structure in place i have right people sitting in the right seats and able to
00:16:33
Speaker
build a good cadence weekly meeting monday planning meeting um you know it companies started to shape up as an organized company to run it you know so it takes about 18 months to two years before you see all these things shipping up okay okay and uh till when does version one of the business continue you you said you've gone through like four versions this is the fifth version so Yeah, so I would call every five years as different versions, right?
Business Evolution and AI Integration
00:17:04
Speaker
1.0 was when we were building a company with web technology. Everybody those days wanted to become ah internet company, ASP, web technology.
00:17:14
Speaker
Then the version 2 came in, everybody wanted to become social, mobile, portal, analytics, cloud. Then version 3 came in where the cloud matured and everything started to go on a cloud and platform thinking.
00:17:27
Speaker
And now later on the platform became into low-code, no-code, domain-centric platforms. ah So you got to mature in cloud and domain-centric.
00:17:39
Speaker
And then finally today is the intelligence economy. So today's version is all about intelligence. um it's I call it intelligence amplification economy.
00:17:49
Speaker
The book which I wrote about six years back is all about digital operating model because everybody wanted to mature in digital journey. ah and In that book, I have written about that after intelligence economy, it's going to be experience economy.
00:18:04
Speaker
It's going to be purpose-driven economy. So all these things are different eras are going to come in in future. But right now, let's live the intelligence and AI economy, which is driving everything around us. yeah So that has been the journey for Salker.
00:18:19
Speaker
Okay. The... like from an organizational perspective what were the major milestones like did you uh narrow down the focus uh or was it like across multiple domains and uh what kind of maturity stages the organization must have gone through as well certain important milestones just take me through those yeah great question um so i would say the Two parts of our organization. So DECID, first DECID, we were like a technology service provider. So we were selling technologies like Portal, Vignette, Liferay, IBM WebSphere, SharePoint, you know, Portal, Enterprise Service Bus, Enterprise Architecture Assessment. So being a technology and, you know, driven company,
00:19:10
Speaker
Later on, we started to structure the company on a domain-centric, business-centric. So we focused on finance and payment domain, insurance on property casualty and life insurance, then manufacturing, retail logistics, higher education, healthcare, high tech.
00:19:28
Speaker
So that's how our domain started to evolve. And that happened because we had a consultative approach to look at every business problem. Because when we go to a client, our lens of looking at the problem was from Dr. Sangel.
00:19:43
Speaker
We were not just selling the drugs to them. We were just looking at problem definitions, asking the right questions, diagnosing the right problems. And if you have industry domain knowledge,
00:19:55
Speaker
you are able to do the diagnosis better. So we changed our company direction based on domain approach and that helped us. That was one of the reason we launched our second company, Culinary Digital, because when we went to one of the food company, we looked at it as a doctor's angle and said that you don't need to buy big yeah ERP product, you need to build a product.
00:20:20
Speaker
It will help you scale your growth in the business. And that's how Culinary Digital shaped up. And we launched the Culinary Suite product, which is the SaaS software ARR-based running business you know for us.
00:20:32
Speaker
I'm very, very happy that that was another significant milestone which happened. Another milestone is... in 2012 11 12 we built our own campus in hingevary that's a great milestone we started building um you know multiple platforms we started to build so we were innovators so we built supply day product for supply chain industry glidoscope for higher education industry later on we built fulcrum one which is a low-code no-code development
00:21:05
Speaker
And very recently, in the last three years, we have built FD Rise AI platform. So because we have been platform development, product development company, along with the software services business.
00:21:17
Speaker
So these accelerators have been the a value creator for our customers, right? So as you are in the journey of growing your business, you got to be differentiating yourself in the market and differentiation came through this maturity of those products. So whenever we launch the product that has been milestone, great milestones.
00:21:39
Speaker
um There were other sub-milestones like we expanded in Europe, we had office in near Manchester in Huddersfield, came close to London. I think being in London gave us an exposure to bigger market.
00:21:53
Speaker
ah We opened up office in Finland and that also gave us a new dimension of growth for us. um Also, going into 2014 in Brazil and Argentina was another significant milestone. So now we could lure local software development team from those markets. So that's so a significant development. And then other milestones are like we started to launch our own tech exchange and Tandem.
00:22:23
Speaker
So Tandem has been our annual event, which we do every year. So and that creates a yearly milestone, but 10 years, 15 years, 20 years has been the significant year.
00:22:35
Speaker
Tech Exchange is the big event which we host in New York City. We invite 200 to 300 CIOs. So that's a very big milestone. Now we have built our own brand.
00:22:45
Speaker
So we are able to attract the partners to you know demonstrate their product and multiple industry leaders to come and speak at our Black Song. So that's a larger Black Song we have created now.
00:22:58
Speaker
Okay, fascinating. I'm going to zoom into some of these. um ah Let's start with ah the core business, which is as a technology services ah is what you are selling.
00:23:10
Speaker
um So this is the same business which like say an Infosys or a Wipro would be in. like good They would be your larger competitors. Correct, because landscape and ecosystem is so big. You have large players like India-based offshore company, and you also have a local company in the US and Europe who are consulting and outsourcing company. They are also our competition, but the real competition is with the mid-tier companies in this space who are a strong value creator.
00:23:42
Speaker
with a local time zone delivery model along with a global model right so if mid-sized company has a latin america india delivery center has built a product has a consultative approach of building an assessment method studio angle not every mid-sized company has a model like that right so then the market starts to narrow down And in that space, you will find only four or five good players with whom we will end up competing with.
00:24:15
Speaker
Right. And end of the day, some of them gets acquired, merged. So the market again shrink and then the new players emerge. Right. So while it sounds like a very big market, in reality for us, the market becomes narrower and you've got to play your game in that.
00:24:32
Speaker
Because everybody plays soccer, whether you play soccer, eight by eight field or four by four field. Yeah. ah I'm not clear why an Infosys is not someone that you see as a direct competitor because if you're bidding for a project, you're saying that some of those projects might not be big enough for them.
00:24:52
Speaker
ah Therefore, you don't see them as a competitor because the… Things you spoke about like accelerator, time zone delivery, I'm sure an emphasis would have the ability and the intellectual property in-house and to compete with you on those things.
00:25:08
Speaker
So is it the ticket size that you operate in a ticket size which they don't operate in? Yeah. So remember I told you in the beginning beginning, everything is derived by opportunity, right? So size of opportunities. So if we are attacking and there are two types of opportunity from the same customers, which Infosys attacking and we are attacking. Say, let's pick $20 billion dollars company.
00:25:29
Speaker
So that $20 billion dollars company has a ticket of hiring 2,000 employees in BPO and IT application support services. I think Infosys is a better suitable partner for that company to do 2,000 people. We are 1,500 people, so we can't build 2,000 people capacity for them.
00:25:49
Speaker
We will get disqualified. So there's no point competing with them when a client is looking for 2,000 people. people However, if the same customer is building a digital technology and digital platform, which requires to only 100 people, 300 people, 200 people with innovative platform, with a better consultative approach, with a deeper deeper domain knowledge, I think we can give run for money to many large companies, right?
00:26:16
Speaker
Because that's the area we are very good at. We have gone deeper. We have studio approach, AI approach, agility, speed, culture, and customers can sense those things better from us than larger company.
00:26:33
Speaker
And that's where we score and we do compete with them on those pace. But that market share is
Specialization and Competitive Advantage
00:26:39
Speaker
different. They do compete with us on those pace, but for them might be a smaller market share where their larger market share is all this 30, $40 billion dollars deal size with large scale people. That's how, you know, model could be slightly different. But we do compete with them. But there are many other players in midsize who are...
00:27:00
Speaker
eyeing for our line of the business, our size of the pie of the business and we compete ah head on with those kind of players more frequently than in process of and the world. So essentially it's about showing specialization which is why it's important to go after clients in ah in a specific lane and stick to your lane because then those clients are going to see you more favorably than IT giant like Infosys because you are specialized in that domain. So you would ah understand and bring more cross-industry knowledge and stuff like that. Okay.
00:27:33
Speaker
And you have a story, winning a story in those large accounts. So you can replicate that and large companies buy those stories. They feel that we're looking for a smaller company and better value creator.
00:27:44
Speaker
And I think you fit in those bills better than some of the large company. And, you know, we're looking for that kind of partner. So there are some large companies looking for a value partner like us. And there is a big market for for all of us in that space.
00:27:58
Speaker
Right, right. A large company would prefer a smaller vendor when they want more agility and attention. Like with an emphasis, you may not get as much as agility and attention, which a smaller vendor would give you. Okay, understood. absolute You ah spoke of how and your value proposition to a client are the the platforms and accelerators. What are these things? you You said studio, platform, product, accelerator.
00:28:26
Speaker
What is the difference between the four of these and how are they a value proposition for the client? Absolutely. See, today when the customers are building a modern applications, right, everybody wants to be relevant in the market.
00:28:40
Speaker
Nobody wants to get disrupted. So for the customers to be relevant in the market, they come to us to build a new applications, new platform for them. When you build a platform, there are three things customers are looking for.
00:28:54
Speaker
Data, modern application logic and how the experience layers are built on the top of it. So these three layers define the modernization of the customers' businesses, right?
00:29:07
Speaker
Now, when you look at the experience layer, Engineers who are doing the programming cannot look at it from the studio and experience mindset. So you've got to have a studio lab and a studio experts who can tell how the consumers are using those applications, reacting to those applications.
00:29:25
Speaker
That's very, very important. Now, today's world, AI and experience are very important. So you need to look at every problem at the lens of AI because AI can solve some complex problem in two steps, then traditional way of doing it in eight steps or 10 steps, right?
00:29:43
Speaker
So we marry AI and studio angle to look at any business problem. That's when we can be a value creator, right? So remember I told you need versus want. So we need to focus on delivering need, not the want, right? So the AI and studio helps there.
00:30:00
Speaker
So, studio is essentially ah a team of people who specialize more on the design side ah would be called a studio. Absolutely. Yes. They have deep expertise and so they can look at ensuring great user experience for whatever is being built.
00:30:18
Speaker
Absolutely. Absolutely. That's what they do every day. They know the new tools, technologies. They are expert in UI, UX. And, you know, there is a different level of maturity around here on that space. And you've got to be deep into it. Right.
00:30:31
Speaker
And then when it comes to application logic layer, that's where the platform accelerator comes into picture, like low-code, no-code development platform, which can further accelerate the software development lifecycle. So if you take six months, can you do it in three months with, you know, those application accelerator? Right.
00:30:47
Speaker
And finally, the data structure, like your ability to integrate with the data, organize the data in the systematic manner so that the applications can consume it and deliver the value. And once you set this structure in place, you can hold the customer's heart with you and they can stay as your IT partner for a long term. They are not going to let you build only one platform, but they will be your platform partner.
00:31:12
Speaker
There will be platform effects, I call it, the multiple platforms they will ask you to build. And then you continue to be a digital engagement partner. It's not an app.
00:31:23
Speaker
There is a difference between application service provider partnership versus platform effect application partner provider. Right. So there you stay on a long term building platforms for them, modernizing it and supporting it than just doing a simple application support for them.
00:31:41
Speaker
OK, OK. um How is your... ah ah offering priced is it on a man hour basis that this is a project which needs 10,000 man hours or is it priced on ah the cost of building this platform is x dollars
00:32:02
Speaker
So there are culinary digital has a different pricing because it's a ARR business.
AI Models and Pricing Strategies
00:32:08
Speaker
That's a saas business or SaaS business. So it's a SaaS pricing. On a Fulcrum Digital, it has a three level of pricing.
00:32:15
Speaker
Nowadays, we are doing some projects because of AI on outcome based. So we'll roll out an AI agent and we will link it with the customer's productivity. And based on those productivity and metrics, we have a pricing model built in, right?
00:32:30
Speaker
But that's not a very common one. Very few customers we offer that based on the scenarios. The second one is outcome-based pricing. So they define the project scope and then there are well-defined ah goals and acceptance criteria.
00:32:45
Speaker
And if you meet the acceptance criteria, you get X dollars. And otherwise, if you exceed the acceptance criteria, they charge you, they pay you 5%, 10% additional money to deliver on the project. But if you don't deliver it, they have a penalty clause as well, right?
00:32:58
Speaker
So that's the outcome-based ah delivery model. So that's a good chunk of business we do inside our pricing. And another one is when the client defines the scope of the work, we manage the project, build the team,
00:33:11
Speaker
Project is managed and governed by us, but we give the transparency of the rate card structure to the client and client is willing to pay on a agile and a scrum basis. Each sprint team takes like 20 people, a project manager, developers, BA, QA.
00:33:27
Speaker
So you build the team, charge them on a TNM basis, but project responsibilities with you. So that's the another third model. So... um outcome based um then performance based totally and then third is tnm based project in degree which is a bigger uh piece of the pie like i Right now, i would say outcome-based. The very neck-to-neck is a TNM and outcome. And the percentage of result-based and AI agent which delivers the outcome and you get a percentage of that outcome, that percentage is slowly accelerating. I'm hoping that number to grow bigger in the period of time.
00:34:10
Speaker
But that's the last to small piece. Okay. Okay. I guess the the legacy way of pricing has been the the time-based, the T&M approach ah for this industry.
00:34:22
Speaker
Yeah, it has evolved. Earlier, it used to be the waterfall model. So many projects used to fail with a fixed price model. Then came the TNM model because on the agility, people who started to and understand a Scrum-based model because that gave the transparency and short-term value creation. You can see what's getting built, believe it before you give them more funding. So because of transparency came through Scrum, so people started following more NMR TNM model, right?
00:34:52
Speaker
But in TNM also, they started oh ah demanding that you need to be committed to all the outcome of the project, right? now So acceptance criteria and thought those things started to mature.
00:35:05
Speaker
Okay. ah So, you know, my question then is that if you are creating an accelerator which shrinks the time by a third, you said instead of six months, you can do something in two months because you have an accelerator and there is more...
00:35:20
Speaker
stuff which can be built without writing code because you already written the code for it ah for another client or things like that uh so doesn't that reduce your revenue potential no but you do charge them for using your accelerator right it's not free right so while you know if it is taking say a thousand um person mandates to build a project, right?
00:35:49
Speaker
And out of that, I'm saying that now, instead of 1000 mandates, i'm going to build you in 500 mandates. So customer is not paying for me 500 mandates, customer is paying me maybe another 300 mandates in a license pricing model.
00:36:03
Speaker
over the period of two to three years. So I'm not getting enough money right away from them. Customer is also happy. They are saving 50% and they are paying me remaining 20%, 30% over the period of next two to three years because it's a license I have given it to them.
00:36:18
Speaker
Right. OK. So their cost of ownersship ownership is a spread out. They feel that our neck is on the line and we are also having a skin in the game. And that's when we are doing the pricing model like that. So they know that we believe in our accelerator.
00:36:34
Speaker
We will license it. We'll only get paid when the project goes into production. So there are two ways of getting value of accelerator, you know, confidence building to the customer.
00:36:45
Speaker
reduction of the price and then partnership over the period of time for three years. ah How much of your revenue is license-based? um If you look at the combined companies, I think overall in our group companies, um I would say somewhere around 30% will be a license-based price involved.
00:37:08
Speaker
Okay. Okay. Okay. So, which is also like and a good indicator to understand that you've invested a lot in creating these platforms. If 30% is a very sizable chunk, so ah you must But I'm using my culinary digital. If you just use Fulcrum Digital, I would say 10% to 12% will be licensed business in Fulcrum Digital. But if you use culinary bit a digital in a group company, then the percentage goes higher.
00:37:33
Speaker
Okay, okay. Now, you know, this business of ah selling technology services ah or IT services is ah like, you know, AI is kind of going to put pressure on ah the the regular selling time and, you know, that pricing by the mandates because now a lot of things would be much faster with ai a lot of companies might be able to offer products which were being offered as services earlier because thanks to ah ai you can offer a more like the reason for a company to go for a service i feel is typically because they want something which is very custom built ah but thanks to ai uh
00:38:25
Speaker
Customization is just a few prompts away, right? So wouldn't that like kind of put pressure on the services market? Absolutely. you know And so every market changes, brings a mindset change, cultural mindset change. I think that's what we need to adopt before even even adopting the AI fabric and LLM and in future ah quantum and things like that. And few people are still resistant to it, but they may be forced.
00:38:54
Speaker
Some people are leader. They have an advantage. you know They have adopted it ah They have failed. So but that failure is a maturity for them. So certainly they will, you know, gain benefit out of it.
00:39:07
Speaker
And some are moderate, right? So moderate will still grow, but very conservative growth for them. Right. So there are three layers of people I see, you know, um and end of the day, what's going to happen is choices are getting created.
00:39:21
Speaker
You can still continue to build your custom AI. You can buy a ready-made AI agent which can solve your silo problems. It's up to you to design your organization structure.
00:39:33
Speaker
Tomorrow you want to build everything silo AI agent, but then you don't have a control on those AI agents. What if those agents become autonomous AI agent? They started to learn about your business faster than you can tell them what your business is all about.
00:39:47
Speaker
Now you have a thousand AI agents bought by thousand different cal vendors and they are all autonomous. Now you have created a recipe for disaster. Now don't have governance model. That doesn't sound realistic to be honest.
00:40:00
Speaker
but It will happen because what happens is ah you you go to a database vendor, right? Or you ServiceNow or Salesforce.
00:40:12
Speaker
Everyone has their own 100 AI agent. and Oh, you want to replace your people? have an AI agent for it. You want to replace the other department? I have an AI agent. Who do you pick?
00:40:23
Speaker
You want to pick Databricks? You want to pick ServiceNow? You want to pick Salesforce, HubSpot? every vendor and then you have a small vendors also existing ready to deploy the ai agent what are your choices and then you have your smart it department telling i have my own homegrown ai agents so who do you listen to but but right so then you compete that time of the day so the best thing i feel like the governance framework is equally important people should good ceos and cfos should create a governance responsible ai framework
00:40:59
Speaker
Who is developing it? Which product should fit in for what role? And they should build a strategy for it. These things must be homegrown. It's a fabric. You need to design shirts, T-shirt, all internally.
00:41:11
Speaker
These commodity products, you don't need to build internally. you There are many vendors. You buy it, use it, release it. When you release it it's a replaceable AI agent.
00:41:22
Speaker
But if it is a dependable AI agent, you are stuck with that. These things people need to start thinking through. And then there are new ecosystems developing in AI with model control protocols and APIs.
00:41:38
Speaker
So any data And any AI agent can start talking to anything right now. And that is the bigger market which you will see in next one to two years. So two things will evolve in next one or two years. Chaos.
00:41:51
Speaker
To solve that chaos, you need interoperability of the AI agents. Right. And people who are a smart player, they will start to interoperate smartly and put under the governance framework.
00:42:05
Speaker
And then the second area as as area which will emerge is autonomous AI agent. The agents which will learn by itself, mature by itself, and that will surprise you with outcome.
00:42:18
Speaker
Today, it's not happening, but it's likely to happen. More and more players are maturing their LLMs with a reasoning model, what OpenAI and Chep&I and all launched. On the top of reasoning model, the memory models are coming in.
00:42:32
Speaker
The short-term memory, the long-term memories, the episodic memories. On all these memories, different brain senses are going to be added on the top of it. And that means now we can use LLM for more intelligent work.
00:42:47
Speaker
Okay. at home I want to zoom in on a couple of things which you spoke of. um but what ah ah mccp Model MCP, model control protocol is the full form? Sorry.
00:43:02
Speaker
MCP. What is MCP? What is model control protocol? You know, ah in the old, in the days of networking protocol, right, everybody started to bring routers, hub, switches.
00:43:14
Speaker
So people start to communicate with each other with TCPIP protocol, transfer control protocol, right? So then you can connect one device to another device with certain protocol. So data can be exchanged.
00:43:27
Speaker
You can build another hardware products, but all of you have to adhere to certain protocol to communicate with each other, right? That was on the network device. The same concept is emerging, similar concept, not same, similar concept in different models are there.
00:43:42
Speaker
Every like chat, OpenAI has a model. um you know cloud has a model and then the different AI agents are getting built on multiple models so these are different models and AI agents so they need to follow model control protocol so that they can have a better handshake with each other right Otherwise you will be running it in a silo format.
00:44:06
Speaker
Or you will have a very challenging way to integrate with that model. Then cost of integrating is very high. You have a security risk with that model. But if it is a model control protocol, your securities are validated, your data is protected, you have ease to integrate.
00:44:25
Speaker
So those kind of phenomena is growing. You can call it a model API economy. you know that is going to become very powerful because there will be hodgepodge of AI agents, hodgepodge of models and how they communicate with each other is the bigger area to solve.
00:44:42
Speaker
so Okay, okay. So MCP is a protocol for how the information is handed off, security, ah guardrails, all of those things would be defined within MCP in terms of how AI agents interact with each each other.
00:44:57
Speaker
I would say how AI agents operate internally, that's the protocol, right? And connecting with the database. So every agent has to ingest the data. So ingestion of that data in that AI agent is happening through model control protocol.
00:45:12
Speaker
And this is a ah like an industry-wide initiative like Google and Antropli. Absolutely. Everybody. It's right now most popular. Maybe things will change. But as of now, everyone is riding that wave and maturing on this space.
00:45:27
Speaker
Okay, okay, okay. Interesting. um You said that companies have to build a governance framework for using AI agents effectively. They have to proactively build a governance framework instead of randomly choosing which AI agent and they want to use.
00:45:43
Speaker
oh What would a governance framework look like? Like, what do you mean by that?
00:45:50
Speaker
So, ah first of all, all these models which you consume need to be understood bare, where the data is residing and how the data is getting ingested in the AI.
00:46:01
Speaker
It should be within your firewall. If you're running a business and you have a customer data, employee data, you cannot be in a public domain, right? so So whichever model you define, you need to govern that, right? Bring it into secure platform. Like you live in a house,
00:46:15
Speaker
You have a door, you have a lock, you need to make sure it's inside the house, right? That's number one. Number two, in the governance framework, who is using what purpose these LLMs, right?
00:46:26
Speaker
So when you run a company, you have a CEO, CFO, different departments, vice presidents, managers, frontline workers, based on the roles and the departments you have given them the access.
00:46:40
Speaker
The similar way, when the AI agents are built, you need to define the protocol for what purpose these AI agents and it's restricted. Otherwise, if you give a free hand to a frontline worker,
00:46:52
Speaker
can go to the CEO's cabin and pull all the files, which is not required for him to pull the file, but he's an AI agent, right? So, you know, you have just not defined the guardrails. So he doesn't, the AI agent doesn't know that I'm not supposed to go to, you know, I have an access to those files.
00:47:08
Speaker
And if some employees asking me the information, I'm going to pass it on to them, right? So you got to secure those kind of a content because LLM is a trained content. AI agent can pick any data and get trained on.
00:47:22
Speaker
So anybody's asking or prompting a question, they will answer anything about your company. What is the CEO salary? How many leaves he has taken? Now, these informations are not supposed to be for those ai agents, right? So you got to define the data structure And based on that data structures, you need to ingest it. And that's why in the businesses, it's no longer a nomenclature of large language model.
00:47:53
Speaker
It becomes a small language model. So you need to start operating in a small language model to have a better governance framework right so so one is inside the house the second is a small language model with a small data sets or predefined data sets with the proper defined data strategy and a structure so that's number one number two number three is how the data is getting synchronized you know how the api connections happening with the data sets together and then the access protocol, you know, who is using it.
00:48:29
Speaker
And then you need to give the access to agent and super agent. So you can keep defining the term that have employees and have a manager and have a director. So based on that, you can call the agent's name.
00:48:42
Speaker
right You will see a lot of people calling different different names in the organization for AI agent. If it is just doing the worker's job, AI agent, doing a manager's job, director's agent, super agent, manager agent, copilot agent, you can keep defining any term you want.
00:49:00
Speaker
But tomorrow there will be term for the CEO agent. There will be co-CEO as an AI agent. There will be co-Akshay doing podcast as an AI agent. gen general So for every role, there will be an AI agent. Just imagine another guy sitting next to you telling you, you need to ask me three more questions that will complete the podcast in this direction.
00:49:21
Speaker
right So you have a co-pilot guiding
AI Agent Development and Governance
00:49:24
Speaker
you. You are still doing the interview, but you are getting intelligence co-piloting through you yeah okay fascinating uh how are you preparing for this agentic future at falcom digital yeah we have a great team internally so they helped us launch the platform rise in 2023
00:49:46
Speaker
That was 1.0. In 2024, before industry started calling agentic, we were already added in the market with 2.0. So what is this 1.0 2.0? When we were 1.0, we had less than 10 agents with us.
00:49:58
Speaker
When we are 2.0,
00:50:01
Speaker
when we are two point zero we have more than three four hundred agents with us for different industries and operating as we are moving to 3.0 we are maturing the quality of our ai agent becoming more autonomous deeper solving more complex problem with the customers and many ecosystems comes together to build the ai agent so many people think that i will just get the yeah ai agent and my job is done it doesn't work that way you need to bring ai agent you need to create some automations around it and then you need to configure it connect the data sets train it you need to do some more programming on the top of it then finally that complex ai agent operates you know it's not just a plug and play you bring the ai agent and it starts working you know
00:50:50
Speaker
Give me an example of a complex AI agent or a complex use case in which an AI agent is ah deployed. So complex scenario is, um say, um two things, right? One is, let's look at the insurance sales agent, right?
00:51:10
Speaker
So if your insurance sales agent is going and selling insurance policy in India, for an example, next, know, Indian market context, I will give it to you. So if there is a life insurance sales agent selling the life insurance product, there will be multiple agents working together, right?
00:51:26
Speaker
to help him solve the ah sell the life insurance. The first agent is doing a planning exercise. So it's a planning agent, helping plan all the banks, profile of the customers, what pitch he has to make, schedule his calendar, um all that thing. So there is an agent who is just operating to help him plan the business.
00:51:48
Speaker
um calendar for him whom to talk whom to drop out from the list because he must be just running around going and meeting people but he's wasting his time and a good ai agent can tell him this customer doesn't have a budget you rather go and meet with this guy it will give you more money so there's a planning agent right once you are planning agent then you have us ai enable AI agent sales. Now he is sitting in front of the customer and he's asking all the questions and there is a social listening skills.
00:52:19
Speaker
So through that, AI agent is listening to your conversation as well as it has a data pre-built in to highlight to you what you missed communicating to the prospect. So prospect asked the question, but you did not satisfy him with the answer.
00:52:33
Speaker
So that's your co-pilot AI agent selling it next to you. So that's a different agent, right? And it has multiple tools built into it, right? It's a social listening skills, delivering the content on a real-time basis to the guy, and and he has to click, and then it gives some more questions to him.
00:52:51
Speaker
So that requires multiple data source. so And then finally, Next agent is signing up the agent, right? So now you already sold it. You need to complete the paperwork.
00:53:01
Speaker
So you have an AI agent which has got everything built in. Make sure it validates, checks everything that you have got this form signed, KYC done, blah, blah, blah done. Check the PAN card. Everything looks good, ready to sign. The AI agent helps you complete the signing process.
00:53:18
Speaker
Okay. Um, I remember seeing, uh, a couple of product companies building these kinds of sales agents for BFSI, um, What would make a company chose an agent from a product business, a SaaS business, which is specializing only in sales agent versus an agent from a company like yours, which is ah like an IT services company, which is and now also building custom agents?
00:53:49
Speaker
You know, what is the way in which a buyer thinks about these two options? Yes, good question. So there are many players who are just building one particular industry silo agent and they are really good if you find that vendor in what but one particular industry and they are only good at doing one agent job.
00:54:09
Speaker
I think they will do a better job than building a generic AI fabric and solving multiple AI agent work, right? We are a farmer. We create a garden of AI agent. We are not just going to create one AI agent.
00:54:23
Speaker
That is called, in our definition, we call it agentic journey. So if customer is buying one agent and calling agentic journey, that's a false statement. Agentic journey is ability to create multiple AI agents for different business problems. So we are a partner and we will be the right partner and a good choice.
00:54:42
Speaker
But if the customer's business problem is very clearly Well-defined for an example other day one friend of mine showed his AI agent which is into restaurant and event management business.
00:54:55
Speaker
So instead of a salesperson making a call and booking the space and giving a quote and all that or showing the space and giving a quote, there is an AI sales agent who will make a call and book the space, right?
00:55:08
Speaker
And that person only wants to focus in his restaurant business. And I was like telling him, hey, why don't you give your agent for some other industry? He said, no, We just understand this business restaurant. We need to stay focused on it.
00:55:20
Speaker
And I respected the view and the vision for that business. So there are companies who are going to stay in their swim lane in AI agent. So if tomorrow I have to build an AI agent for ah restaurant event management, I might refer somebody to go and buy that product because they are that's what they do every day for living, right?
00:55:39
Speaker
And then with my MCP protocol and API interface, I can still integrate with my agentic journey, right? I don't want to go there, but that's my commodity. I can use it, release it, right? I'm not building my whole differentiating business model on the top of AI agent, but anything which is my core, I would write rather like to capitalize so on my agentic framework, you know.
00:56:02
Speaker
and Okay. Okay. ah What is an AI agent? I should have probably asked you this earlier on, but... ah I'll tell you my layman understanding and, you know, I want you to help me get a more nuanced understanding.
00:56:16
Speaker
ah i see like an AI agent as you have a, like say an open AI API or a Gemini API through which you are,
00:56:27
Speaker
doing calls and those calls are either to give back data, for example, a WhatsApp, ah someone, an agent which is doing a sale of a product through a WhatsApp conversation. So every time the customer is saying something that goes to the AI, the LLM that you're using, which processes and gives an answer, the answer goes back to the customer and that whole journey is happening through this API connection, whatever. Is that what an agent is?
00:56:56
Speaker
I don't think so. So agent is this. um Agent is um like you call some employee in your company, you're a marketing department or finance or delivery, and you tell an employee, I want you to complete this task.
00:57:12
Speaker
So the definition of agent is you define the task first, what that task is, right? And then tell the agent, you need to go, you know what you have to do, you know, this is a defined task and I want this outcome.
00:57:27
Speaker
So you have defined the task, you have given the outcome. Agent job is to go and follow those 10 steps, 20 steps, interact with different groups, consolidate all the information, um and then you define that engine and it will operate and deliver that task to you.
00:57:43
Speaker
So that is called an agent, right? So defining the task, and defining the acceptance criteria is the job of an agent. So it's like you give a task to some employee and then after three days you come back and say, hey, show me the result.
00:57:58
Speaker
The guy has worked on a PowerPoint presentation. He has looked at all your numbers and you are going for a proposal presentation. He has compiled everything for you. um He has all the research notes. He tells you with a citation where he does found all the information for you.
00:58:13
Speaker
So it has taken him three days. Now with AI agent, in one click, in five minutes, you get an answer. But the only difference is now you are not waiting for that employee to follow your task and give you the result.
00:58:26
Speaker
AI agent is giving you and now you are saying, no, I like this. I don't like that. I like this. So now you change. and Now you tell that employee agent, I'm good with it 70%, but you need to change 30%. My style is this.
00:58:38
Speaker
Now they will follow your instructions and then deliver you closer to your expectation. So that's in my definition is the AI agentic job. right Tomorrow, the agentic job could be a job of one person or it could be a job of group of people.
00:58:52
Speaker
Because in marketing, you need somebody to do a content management, somebody to do graphics creation, somebody to do the video creation. So it's a job of three people. So ai one AI agent will find the resources and create those tasks and is still deliver to you.
00:59:08
Speaker
So you still created one AI agent, but it is doing a multiple skills in that task. you know Okay. ah like There is this deep research feature in most LLMs.
00:59:23
Speaker
that That is also like an agent then where you you give it a problem, and then it goes out, does ah a Google search or whatever and collates the information and then makes a research report on it. That is somewhat of an agent experience.
00:59:35
Speaker
Yes, you can call that as an agent because end of the day and what it does is because it has a reasoning ability because now with lot of reasoning, it starts to look at your prompt and start to reason. And based on that reasoning, it starts to go to different sources and cites all those sites like, okay, I'm working, wait for two minutes, I'm working on 300 sites for you.
00:59:58
Speaker
It takes time for 300 sites to scan through, analyze it and then present the data. Certainly it's a small micro AI agent if you want to call that as a definition for the agent of course based on your prompting it can do the job but You know, I would go one step ahead in the AI agent in today's world,
AI Agent Types and Future Prospects
01:00:18
Speaker
right? Because agents are vertical AI agents or horizontal ai agents.
01:00:23
Speaker
I call it a vertical AI agents, which brings customer productivity, brings revenue to your company. Horizontal AI agents are employees' productivity improvement. If employees take 10 days to do it, if they can do it in two hours, that's a horizontal ai agent, right? And there's one example where financial advisors, they want to send to their clients financial reports.
01:00:48
Speaker
AI agent can create all those reports in five minutes and give it to you to submit it to your client. If manually you have to go, you have to go 10 sources, look at the historical data, look at the template, look at the last queries what customer asked, customize, prepare the report. It takes them two to three days, four days, whatever, right?
01:01:08
Speaker
Now, a agents are doing it. So I would suggest those are... more agentic kind of an approach those are employees productivity so i'll call them under horizontal ai agent but the insurance so ai agent which i talked to you about improving the sales those are called vertical ai agents so if you look at the company you will have a cluster of horizontal ai agent and cluster of vertical ai agents and that's the agentic journey you are on
01:01:37
Speaker
why is the term vertical and horizontal used for these two uh i would have thought vertical means like deeper but narrower specialization and horizontal would mean general all-purpose able to do multiple tasks but uh from the example you gave me like preparing reports which you send a financial aid advisor sends sounds like deeper narrower specialization but you calling you're calling that horizontal uh what is the the logic behind that my understanding is on x-axis you put in employees on a y-axis you put revenue so when your employees are working on a product and services it's impacting the revenue
01:02:24
Speaker
and basically customers. So on the Y axis you put a customer, on the X axis you put a employees. So as employees continue to work in a highly productive manner, it's impact your customer's revenue growth and impacts your revenue.
01:02:38
Speaker
So it's that's why the Y axis of customer gives you a vertical agent. because it's impacting the revenue and why axis is an employee's axis, which gives you employees productivity, which has which you can see. So if you see the growth chart of yours, the business, when you bring more and more AI agents, it should have a hockey stick growth, growing chart. if If you are adding more and more AI agents and if you are declining, then it means your revenue is going down as you add more AI agents. So then it's vertical AI agents is not working well for you.
01:03:10
Speaker
Why can't you have one agent to rule them all?
01:03:15
Speaker
That's an interesting twist in the question. So you are heading towards AGI and singularity and the future of agentic journey. So i want you to think like this. There is a pyramid and there are five layers of agent, right?
01:03:30
Speaker
So one layer is layer one, which used to happen a conversational AI agent. then you have today's ai agent which is vertical horizontal different business function agent tomorrow you will have autonomous ai agent gen Later on, department is operating autonomously, right?
01:03:50
Speaker
And then you have a CEO's which is operating it autonomously. It's coming. But once we reach that stage, it's an AGI stage, right? I already talked about we are not there where agents can improve on its own and we will accept it, right? It will take a year to two years time to mature in that era.
01:04:10
Speaker
It will take another few years for us to mature into departmental autonomous AIHM. For us to accept that marketing department, I can run with one marketing person instead of 10.
01:04:23
Speaker
Or I can run with five marketing person instead of 20, right? So things will change and you require a co-pilot for marketing department. Then eventually you will have a co-CEOs and CFOs as well, right? So then one person percent can do.
01:04:37
Speaker
But end of the day, the way I look at it, in this planet, we have 8 billion people. And what we need workforce is more than 100 billion.
01:04:49
Speaker
You can't have that many workforce. So you need that many AI agents. But those AI agents have to coexist with the governance framework with human. to run it and that's where the economies of scale comes into picture. you are producing more outcome, you will have tomorrow quantum, more computing power, more outcome.
01:05:10
Speaker
So you need more agents. You don't have that many brains and hands on this planet. So you will create more humanoids, you'll create more AI agents and we will learn to coexist with them within the governance and guidelines principle and that's where the future is. you know Okay.
01:05:27
Speaker
A slightly different way of asking the same question is this, that um Why do I need to create an AI agent separately for each use case?
01:05:39
Speaker
Why do I need to create an agent for the financial planners who have to send reports to their clients and a separate agent for ah a salesperson which will help him plan his day?
01:05:52
Speaker
Can't one agent do both? Because, I mean, that is how, as a layman, when I'm talking to a Gemini or a Chad GPT, I can ask it about biology, I can ask it about physics, I can ask it about marketing, I can ask it to generate a social media post and it can do it all.
01:06:11
Speaker
So why is there a need to have independently made agents? Why why can't one agent do everything, have that versatility? Because a chatbot seems to have that versatility.
01:06:24
Speaker
Right. um I would like you to understand the plane analog analogy, right? And at any point of time, there will be about what, 15,000 planes in the sky and they're all flying. yeah But just imagine all those airplanes are like AI agents.
01:06:39
Speaker
They can do the same job, but they are not doing the same job. They are flying. One is flying from Newark to Mumbai. One is flying Newark to Paris, Newark to London. It's a different crew. So you need a different crew, but the plane is the same.
01:06:54
Speaker
It's the same large language model, it's the same plane, but in that plane, you need a different crew, different captain. Otherwise you give the same captain, it will just drive you to the same destination.
01:07:06
Speaker
So as long as you're not changing the plane, so your definition that can I have one AI agent? Yes, it's an LLM, it's a plane, you got a plane, it's an AI agent, but you need a different crew to deliver different outcome.
01:07:21
Speaker
and that's why a agent difference so when i told you in the beginning mcb protocol data ingestion that says nothing but putting a different crew inside the plane and the moment you put the different crew they have been set with different directions that your direction is london not paris so they will go to paris not london Okay.
01:07:43
Speaker
So essentially, and agent which is making reports for financial planners needs to be trained on how to make those reports. They don't need to learn about how to plan the day of a salesperson. And if you try and use one agent for both, chances are that there will be… it will be chaos, right? Just imagine. Absolutely. the which you have sent it for somewhere and it's landing somewhere else it will be right observable experience right okay understood why would agent providers continue to exist would there come a day where i can write a prompt and generate an agent why would i need to use someone like a falcrum digital to build an agent for me
01:08:26
Speaker
Yeah, absolutely. So those days are coming. By the time you reach to those days, the market demands will be something else and we have to mature that next generation of technology faster than you can adopt, right? That's how we have done it in the last 20 years because we can see the future faster. We can experiment it faster and be ready to deliver up upon it, right?
01:08:48
Speaker
So when the market is ready for you to build on your own, I think we'll be ready to build the whole business as an AI agent, right? Not as a person as an AI agent, right? So we'll be able to deliver better value. We'll say that, why are you acquiring companies?
01:09:02
Speaker
I'll give you two agents to be acquired. Because I've created a company for you. Why you are building just the one silo agent inside the company, right? So the market opportunities will create ah for us, like tomorrow quantum will come in into mainstream on a room temperature.
01:09:19
Speaker
It will have a better computing power. It will have an edge computing power, you know, edge AI, quantum AI. So we got to learn those things faster. We got to leverage the space technology, the brain control device technology faster rather than staying confined into the AI era where you are going to mature faster than us anyways.
01:09:39
Speaker
Right. So we got to be the, you know, leapfrog in those areas and be the, you know, guide for you. And that's when you will buy the value from us. Otherwise, all these ai agents are going to become commodity, whether it is everything what you're doing, prompting, smart prompting, you know RAG, maturity in the RAG.
01:09:58
Speaker
All these models are deep seek mixture of experts technology, which is coming in. God knows next release, which will happen in the world in next few months or few years, people will get shocked.
01:10:11
Speaker
Now they'll say, I don't even have to do this kind of prompting. i can Today also you can upload the boarding pass and it knows everything about you just by the boarding pass without ever prompting.
01:10:21
Speaker
It can give you what you're flying from here to here. It means you're looking for a car. It means you're looking for restaurant. It means you're looking for this. You don't even have to ask. and Information is already given to you by just boarding pass uploading. right So just imagine you will sit in a room with your video message for a few minutes and it really knows what you're looking for.
01:10:40
Speaker
You don't even have to have prompt it. Because language model has matured to a level where it can see your body language, it can hear your voice tone, and you just speak.
01:10:51
Speaker
And it does the job for you, right? So that era is also coming. So when other thing comes up, rest of the things becomes commodity. Price goes cheaper. Then there is no value there.
01:11:02
Speaker
And if we continue to exist in that market, then we will be disrupted. So we got to learn newer things faster, better, cheaper, and then be ready to guide our customers here. Okay.
01:11:13
Speaker
um Today, every vendor, small, big, who is in this technology domain, be it like, say, a website developer, whatever, is now offering AI agents...
01:11:27
Speaker
what is the ah way to evaluate the quality of an ai agent because the the ability to build an agent is fairly democratized right you just need an api connection to a large language model so somebody who is sitting in a like somebody who's running a five person ah website development agency is also offering an AI agent and you're also offering an AI agent. How do i judge the difference in quality?
01:11:59
Speaker
Great question. One other very good question you asked. um Because in this mass market, multiple vendors, everybody is in this race. So consistency is the name of the game, right? Because what's happening is your data is not constant. It's changing every day.
01:12:17
Speaker
It's getting updated every day. It's maturing every day. Your AI agent continues to deliver the outcome with the old data sets. Not understanding the new data sets is the first quality check.
01:12:30
Speaker
So as your business continues to get updated, if your AI agent in is continuously delivering you an outcome which is accurate, you will trust that agent. The moment they start delivering 2% less data, your trust is going away.
01:12:47
Speaker
And that is the most important aspect for any AI agent is the consistency of delivering the outcome. if it It has to be 100% accurate.
01:12:57
Speaker
Say for an example, if you have an AI agent for invoice to be generated, you don't want send the invoice to your customer, which is $200, but your agent has generated $175. Will you use it? You will kill it.
01:13:10
Speaker
you will kill it But you you wouldn't know that and this is 98% accurate until you've used it for ah couple of months.
01:13:24
Speaker
Before you buy, what is are there some technical due diligences you can do? Yes. So, AI agent development model… It doesn't take too long to build it, but to put it into production, to ingest the data and test it takes a longer time.
01:13:40
Speaker
So if anybody just trying to roll it out faster and cheaper are set for failure. years So they got to stay back and spend more time in validating and testing it.
01:13:55
Speaker
And once they are 100% sure with all kinds of scenarios in data upgrades, changes in agent, then their trust will be high in trusting those agents. So like if you ask your vendor for ah his testing framework or how will you test this agent? So that could be a good due diligence to do before you buy an agent.
01:14:15
Speaker
Like the amount of investment a vendor is doing in testing the agent, how mature and sophisticated is the process of testing and validating the agent is a good ah indicator on quality.
01:14:29
Speaker
Absolutely. Absolutely. In fact, and we were presented with some scenarios ah in a very commodity market, I will tell you, where you have OCR readers.
01:14:39
Speaker
Everybody uses it. It's like lot of PDF files. AI agent can read all the PDF files and give you the results, right? And you can do query on those PDF files. Just imagine some invoices are in color, some invoices have handwritten notes, some invoices have different fonts, ah different table structure, not one invoice is in similar format.
01:15:03
Speaker
AI agent has to actually understand in whatever format, whatever color, and then segregates all data and information so that when you prompt it you get the right 100% accurate result.
01:15:17
Speaker
But that if your AI agent is not programmed correctly, and if tomorrow your invoices are colored with the red color, right, and it misses to read those content,
01:15:29
Speaker
then when you're prompting it, you're not getting the accurate result. So it's also the way it is programmed, configured, and then the testing. If it's not giving the result, you're going to disqualify the agent.
01:15:41
Speaker
In fact, today, I would say that if if the companies are building hundreds of agents, in my guess, 70% are getting shut down because of failure of meeting the right expectations. they have it Many of the companies I know, they started a Gentic journey about a year back.
01:15:59
Speaker
But if you ask them how many agents are into production, only few will be into production. Then what happens to REST? They said they to kill it because it wasn't working the way they wanted it.
01:16:11
Speaker
Okay. um but When you say program an agent, ah what do you mean by that? like i like Do you literally write some sort of code? Is there a coding language to program an agent or is it mostly like prompt engineering?
01:16:27
Speaker
It's a combination, it's a prompt engineering, it's a Python code. You got to put some, you know, your own machine learning language model on the top of the LLM to deliver the desired outcome, right? Because every task I mentioned to you, if the plane has to fly,
01:16:44
Speaker
plane is the LLM but you have different types of planes you have chat GPT you have cloud deep seek you know Gemini so depending on which plane you want to fly depending on the distance and the purpose but then the crew who comes in that crew has to be given instructions at this time you got to serve the meal this time so that has to be written in Python language So inside that when you are configuring and building the AI agent, you will use the LLMs and you'll consume LLMs, but you will consume it based on your own um language model to define the task, which you can call it a microservices, which you will have to write that for this purpose, this task you have to do.
01:17:30
Speaker
And then it will connect to that specific database. those connections can also be written into python it can be also consumed with some other api protocols but those are the areas where you need some coding to be done for configuration customization for a specific task automation because you need to automate some of the tasks before the ai agent understands the work because you are receiving invoices from 10 email boxes.
01:17:56
Speaker
But AI agent needs to synchronize all the 10 email box. So there has to be some automation script has to be written. So you will write those script and then feed it to AI agent. So it requires few boxes to be...
01:18:10
Speaker
coding to be done, it's just not there. That's why many people fail in AI-agentic journey. Oh, I got the fabric and I can query anything and get the result. And then they don't get the result. That means that's when they fail in AI-agentic journey because they just got the plane but without a crew.
01:18:27
Speaker
Okay. Got it. Interesting. Um, okay. How are, uh, companies pricing AI agents and, ah from both sides, one would be like, say, pure SaaS companies, uh, who's basically just selling an AI agent versus a company like Fulcrum Digital, which is a services business and is also making custom AI agents. So, you know, that off the shelf agent, custom agent, how are these being priced?
01:18:57
Speaker
so if you look at enterprise customers the large corporates right because they want the platform to be secure inside they have a three layers of pricing um what data sets we are consuming for how many users and so there is a user based pricing for the department and business objective so when we give the plane we define the plain job is to go from here to there, right?
01:19:23
Speaker
And for that purpose, we price the per user pricing. Then we have one-time configuration installation pricing because we need to bring everything inside there, firewall, and configure it, run it, test it, you don't deploy it.
01:19:38
Speaker
And then there is a consumption price depending on how many tokens they are using it in their data set structure safety case. So um because when you use all these LLMs, we got to pay them for using their tokens. So there is a token utilization. So there are three ways of pricing when you deal with the enterprise customers.
01:19:55
Speaker
but then when we go to say 500 million dollar company 200 million dollar companies and they are looking for AI agents they don't believe in these things they just tell us very clearly have these two employees and I want to reduce my cost I'm paying whatever quarter million dollar to these two employees and you save me 50% so I'll pay you 100k and don't charge me per user all that pricing just give me a flat core month fee And if I'm able to save 50% cost of my employee through an AI agent and your pricing is simple enough, I'm willing to pay you $3,000 per month,
01:20:33
Speaker
you know, about $4,000 per month, I can get you two agents, two two employees can be replaced with that. And I'm happy. you know, I think in 50K, 100K, I can get my work done from you.
01:20:44
Speaker
Instead of quarter million, I'm saving the money. So they are just looking at it. Remember, I told you horizontal agents. So we have to justify them, employees productivity there, and then they buy. ah Some of the places which I mentioned to you, they want to put the agent into production, they want to see the result and then they want to pay So there it can be a transaction based price.
01:21:08
Speaker
So if you are helping them to transact and then you can charge them on a transaction based price, right? Based on if it has improved the revenue, you can ask them for percentage of the revenue, right? um because now you're helping them grow the business.
01:21:21
Speaker
It could be a small percentage, significant percentage based on what value creation you're doing. So so these are the types of pricing works. Enterprise customers different, small customers different.
01:21:33
Speaker
Because we went with the small customers, all three layer pricing, they got confused. They said, ah, this is too much. And then there are some commodity-based pricing also, like, you know, some of the AI agents which I mentioned to you.
01:21:44
Speaker
They are a very small vendor, very silo. They are very simple price, $300 per month per per user, just buy it. So customers are like, okay, this is just $300. I'll just buy this agent. Let me try it out.
01:21:55
Speaker
But that's for a specific task.
AI Project Roles and Tools
01:21:58
Speaker
Easy to, you know, plug it in and see the result, know. So off-the-shelf agents are basically priced like SaaS, like a ah whatever, like $100, $200, $300 a month.
01:22:10
Speaker
ah You get an agent which is off-the-shelf, not deeply customized, but probably ah very quick to deploy and you can see results very fast.
01:22:21
Speaker
ah And then you spoke of outcome-based pricing, like say, if your customer service is being handled, so then every ticket resolved is payable or some number of ah man hours that you are replacing through AI agents or like there are these...
01:22:39
Speaker
call center alternatives like voice agents which are doing calling so instead of having a call center with humans you have these voice agents which are doing calling so that maybe per call kind of a pricing so these kind of outcome-based pricing models uh is mid-market is basically going for this outcome-based pricing models uh and for enterprise based on yes okay you you defined it very well Okay.
01:23:04
Speaker
ah o You are largely ah pricing your agents in like outcome based or is it more of like the cost plus? Right now it's all going all three because there are some large customers.
01:23:17
Speaker
They want a very enterprise pricing, different strategy. There are smaller customers coming. They want to see the value right away. So they're jumping onto it. And there are um customers who are not convinced, but they want to go on this path and they want to be partnered with you. So then they are outcome-based. so all three are going in.
01:23:36
Speaker
Percentage of outcomes are very small, but percentage of enterprise and small agents are high right now. Okay. What is your mote in this business of delivering agents?
01:23:50
Speaker
Because the large language model is common for everybody. Anybody building an AI agent is using the same LLM, you know, the MCP, all of these are very common things available to everyone. So so how are you building a mote?
01:24:05
Speaker
Yeah, so our RISE platform, if you see, it is the best of the breed of LLMs, right? So we consume all the important and relevant LLMs are part of our FDRISE platform.
01:24:18
Speaker
So it's not LLM, one LLM. So based on your business requirement, We will trigger some different LLM to deliver the job. Right. Based on some other requests, I'll trigger some other LLM to get the job. So we have orchestration engine built inside the RISE, which is seamless based on the business request and the prompt.
01:24:40
Speaker
We trigger it On the top of it, we have our own LLM um built on Python language, which also adds value on the top of LLMs. Having that engine which orchestrate, configures, adds value with own LLM, industry-oriented language model, Python written on the top of it, makes us unique in the small language model rather than large language model.
01:25:05
Speaker
So we take the large language model and we configure into a small language model is the benefit of using RISE platform because then we become powerful LLM for their business because they don't care about what's happening in the world.
01:25:19
Speaker
They need to care about what's happening inside their world with their customer data, with their employees data, with their invoice data, what's happening in 10 other companies invoice data. Why do they care? Right. So we we have an ability to configure LLMs with multiple LLMs.
01:25:36
Speaker
which we call it a multi-model LLMs. So multiple LLMs, multi-model LLMs. So you need to have that orchestrator. And many companies have tried to build the orchestrator, not succeeded. And that's where advantage for us comes into picture.
01:25:51
Speaker
So if you have cracked the code, you have ability to build it, ability to continuously consume the next LLM faster in your platform, it means you are able to deliver the good results to your customer.
01:26:03
Speaker
You have the power of that engine. So if I understand correctly, one mote is orchestration, ah being able to create a strong orchestration engine. engine And the second mote is being able to create SLMs, it's ah small language models, which are yeah which are which give benefit of cost also, or is it just specialization?
01:26:23
Speaker
So one, ah obviously, in a small language model is specialized in a narrower domain, but is it also cheaper to use a small language model rather than an LLM? Yes, a small language model is cheaper as well as it's more relevant to them. Relevance is more important because then they can trust that model.
01:26:41
Speaker
If LLMs sometimes too much of information, it can give some vague information, irrelevant information that can confuse the business. So um is that information important?
01:26:54
Speaker
Important, but not relevant for their business. ah Do you need coders to build AI agents? so What kind of workforce do you need to ah be offering AI agents?
01:27:06
Speaker
Good one. um So when you build a team, you need three players. One is um ai business analyst who is understanding the business requirement very well, right? Remember I gave you a plain analogy.
01:27:19
Speaker
So you have to define from... jfk new york to paris so somebody who knows how to understand that the flight route has been defined from new york to paris right so that's the role of a business analyst right to understand the objective of the agent remember i told you the definition of the agent is you have to give tasks to someone it's not a prompting it's a sequence of tasks which you're defining right so um So that analyst will able to understand that. Then you need a coder who is an AI engineer, Python developers or LLM developers. They need to now write on the top of it.
01:28:03
Speaker
automation, configuration, microservices, so that now LLMs are working as an SLM for your environment. And then finally you need a testing group people, right, who are continuously testing it, betting it. And once you have these three teams in place and certainly project manager to run it, then your project team is complete.
01:28:22
Speaker
to the Okay, fascinating. Do you use AI for testing also? Of course, yeah. so In software development, you know earlier, there are a lot of AI tools now, including Copilot, our own tools and market offer in GitHub, all these tools and third-party products where if you are writing a manual script, you can now write automation script.
01:28:47
Speaker
Even using automation, Selenium, another tool, you can generate, say, 10 use cases in a day. Now with AI tool, you can produce 40 use cases in a day. your productivity just quadruples by using ai tool for testing because your ability to produce more number of use cases test cases are much faster earlier it was manual too much time too many less test case automation tool selenium improve the productivity ai tools are further increasing the out so yeah you know you will able to test it faster cheaper better you know
01:29:26
Speaker
Okay, okay.
Growth Focus and Investment Strategy
01:29:27
Speaker
um Are you also doing any sort of angel investing? No, not yet. um We get pitched by different people.
01:29:37
Speaker
but Some things are exciting. We invested in culinary. We want to focus on one thing. We want to stay focused on Fulcrum Digital. That business is growing very well. 1,500 employees, we want to go to 1,800.
01:29:49
Speaker
Tomorrow, we might have 500 AI agents with 1,800 employees. We are focused on that. um will If there are any adjacency market in our culinary business, there is a good opportunity for us to do angel investment. So more like an entrepreneurial fund and we can fund like four five startups.
01:30:12
Speaker
That's an aspiration. Are we doing it? We are not doing it. But if an opportunity presents, we will do Today, you prefer to plow cash back into the business and instead of taking cash out and doing it and investing. We are continuously building this AI Rice platform. We are investing big. So all our profit is going pumped into...
01:30:31
Speaker
building this ai agent otherwise we feel will become irrelevant right so and today we are enjoying it we are getting so much of attraction in the market when people look at our eyes platform they say other competitors are not showing what you are showing so you are certainly two steps ahead of them right that's because of our investment so then they come to us how much of uh fulcrum digital equity is with you
01:30:55
Speaker
So, yeah, majority is with me. ah My CEO has some percentage, but I'm the majority stakeholder. like Like some ESOPs, other than that, it's yours, basically. Yes, we have employee ESOPs, a small percentage to my CEOs, some ESOPs, and rest of them are majority with me.
01:31:15
Speaker
So, how do you... ah like Like, you know, how do you grow from being ah small it services? Like we gave an example, like a five to ten member person. You could have stayed there, ah but today you are at a hundred million dollar revenue.
01:31:34
Speaker
What are those mental frameworks because of which you are where you are today?
01:31:42
Speaker
You know, we all have read the book or heard about it. What got you here will not get you there. It's a mindset. And it took me also quite some time to even even get onto that journey, right? It's a human nature.
01:31:55
Speaker
What you get but got you there will let you stay there because your mind wants you to stay there. And that's a human nature. So breaking away from that your own self nature that let's not be comfortable, be uncomfortable, and finding that uncomfortable path is the recipe for growth.
01:32:15
Speaker
And it's a discipline, like you go to a gym and if you're lifting 20 pounds every day, your body will not show the result. But if you are changing the weights, improving the reps, you did 20 pounds, now you're doing 25, 35 pounds.
01:32:28
Speaker
You did 15 reps, now you're doing 20, 30 reps, you will see changes. So what you are doing is you are confusing your body muscle, you are challenging yourself for more. and um and you have the mental discipline of going to the gym every day or going for four days in a week so discipline making yourself uncomfortable when you get comfortable and staying relevant because market always makes you irrelevant right so these three things if you keep in your game whatever game you're playing and every game has a number of hours sincerity hard work that's given so you've got to have discipline
01:33:09
Speaker
Don't be irrelevant and always make yourself uncomfortable. The moment you are in comfort zone, you know you you have a recipe for disaster that's coming soon. Give me an example of going out of comfort zone that you personally went through.
01:33:24
Speaker
um So if you are running an account and that customer is just giving you more and more business and you are happy with that customer's business, but then what you are doing is you are making yourself uncomfortable dependent on one customer so few customers okay if tomorrow something goes wrong with the customers your business continuity becomes a risk to you right so while it's a human nature to enjoy the journey be comfortable don't push yourself too hard uncomfortable journey is find another 10 customers of that size find another 20 billion dollar company
01:34:03
Speaker
and make them as an IT partner for you is a daunting task. And it it's it takes a lot of effort. So do you want to go that route or do you want to focus on the same customers and go and meet with them and meet 20 other people in the same customer?
01:34:18
Speaker
It's easier way to do, right? You're comfortable doing it. You know the people. Why do you want to go and meet other 20 customers and build a pipeline? And it's 20 guys will say no to you a hundred times. And then you have to break. So it's a...
01:34:32
Speaker
And then continuously you have to show something new and fresh and it's more challenging. So it's an uncomfortable journey when you want to scale. But then the beautiful thing is now you have a better business, continue to plan. plan You don't have a dependency on one customer. on so Customer concentration is less.
01:34:51
Speaker
ah People will look at your business as a much more robust and solid. Your employees will come come feel comfortable to work because now you have mitigated the risk. So you've got to continuously chase that uncomfortable journey.
01:35:04
Speaker
And then also, you cannot do business always with $20 billion dollars company. You've got to find billion-dollar company, $500 million dollar company, because if something goes wrong with a large company, you have another set of customers who can take care of you when the recession hits. So you need to make your business a recession-proof as well so you can mitigate the risk.
01:35:24
Speaker
by different size of customers inside your business so that's another uncomfortable phase the the sales people and solutioning guy who are used to selling larger deals are not used to selling smaller deals right small size requires smaller delivery capabilities and overheads are smaller if you try to put larger overheads on a smaller deal then it's not profitable for you so it's a different game but you got to start playing that it's a garden in a garden you need all kinds of trees all kind of flowers so you when you're building a business you're building a garden you are not building just one tree two tree and just sitting and watching waiting for mango season to come in but what happens once the mango season is over right fascinating okay um i have this impression that uh
01:36:14
Speaker
Nice guys don't scale their businesses. ah you You need to be willing to be ruthless as a founder to really get to scale. and Do you agree with that?
01:36:26
Speaker
I don't think so. No. Everybody is nice. You know, see, how they come across may not be nice. But intentions are always nice.
01:36:38
Speaker
So what you are seeing is how they presented themselves and how they conducted in that scenario may not be nice. And if they reflect on it and they go back and think, because in that spur of the moment they reacted or they said something.
01:36:52
Speaker
But all these successful business people, they all have a very good intentions. You know, anybody who has become successful, right? I feel like somewhere they also understand how to use fear as a way to get the best out of their employees.
01:37:06
Speaker
Those are wrong tactics. I would suggest it's a maturity model. Some people have matured with inspirational model. ah Some people, you know you know, if you read a book of, I don't want to name it, some of one of the top prominent personality in the world today, they will say smack with your baseball bat and then sit on the negotiation tactic. But that's their art of deal making, right?
01:37:32
Speaker
ah you've you've You've given away the name.
01:37:37
Speaker
So, you you know, ah and and then you you you tend to negotiate different way, but that's how they have become successful. They have their art of dealing, right? They take tough stand.
01:37:48
Speaker
And when you take a tough stand, art of deal making is much better, right? So um many people will believe that, no, I can still get the deal done. It's a deal making, in my view, is a break even point, right?
Deal-Making and Employee Evaluation
01:38:02
Speaker
It's a point where deals can be done. If it is tilted more towards you and other more towards them, it's not fair deal. You know, it's certainly a beneficial deal for one party. That's why people try to negotiate pretty hard and somebody wins big time, somebody loses big time.
01:38:18
Speaker
I don't like that deal. I'm a fair deal making guy. right It should be a balancing point and one should recognize that. One should have an ability to pay and consume and one should have an ability to ask that right price. right And if that's there, then it's a long term sustainable deal.
01:38:36
Speaker
But if it is a twisted deal, it's a short term deal. Someday some guy will realize that my hands are tight and like, you know, some country will realize somebody has been taking away from this country a lot for a long time.
01:38:51
Speaker
Somebody will rise one day and try to negotiate with the world, right? so So things will change. But yeah, deal-making is an art. Everyone has a different style and my style is different. I certainly would like people to respect what my product and services are. If there is no respect, there is no deal.
01:39:12
Speaker
And the respect can be only seen by the right price when you want to offer it. know How do you deal with underperforming employees? and What about you? it's ah It's a new twist to your line of questions. so
01:39:30
Speaker
um Because we have promoted loyalty, integrity in the company, so sometimes, not I will say all the time, there are chances that long time employees or sometimes performance can be compromised because they have been in the company for a long time.
01:39:45
Speaker
Where the new employees, they are on the check, company is demanding more from them, so they they will feel that, okay, company is expecting more accountability. from me where other employees have been trusted and there could be a slack from their side it's not necessary that all the cases but in some cases it can happen the way I have started to categorize now and we seen our company is we put them in a magic quadrant of a B plus B minus and C on x-axis we put culture rating
01:40:22
Speaker
on a y-axis we put performance rating. So employees are rated 1 to 5, employees are rated on culture 1 to 5. So if culture employees are 4, maybe performance is 3.
01:40:35
Speaker
You need to coach that employee, the employee will fall from B minus to B plus with a little bit coaching. right But if the employee culture fitment is 3,
01:40:46
Speaker
three And performance is four. Even if you keep coaching them, they will still stay in a B b minus category. And for us to convert b plus to A is much easier. means culturally they are aligned with better training, coaching, clarity, motivation, inspirations. They can move from B plus to A faster, where you keep inspiring B minus people who are not culturally aligned with your company.
01:41:12
Speaker
Moving into B plus is a daunting task. Forget even thinking about A. And they would rather to stay back in B-minus because they are thinking about themselves more than thinking about the customers and the company more.
01:41:23
Speaker
right Right. And so you got to identify that and... build your business model whether they are the long-term employees slack in your company lack of accountability you've got to identify them and just start giving them the feedback and nowadays give them continuous feedback don't wait for six months appraisal if you don't like it they're slacking just give them in two weeks tell them that listen step up you are feeling here people like to get that feedback from their bosses immediately
01:41:55
Speaker
it Give me a practical example of judging culture. Like what what do you judge when you're judging someone on culture? um the What I was just saying, like sometimes people are self-centered. They're only thinking about themselves.
01:42:10
Speaker
They are like, okay now I got more portfolio in my hand and now company has to depend on me. So let me twist the arm, get more advantage for my personal benefit.
01:42:22
Speaker
That means profit will go down. That means customer's delivery can be compromised. That means if power is not given to me, if the title is not given to me, it means in a short term,
01:42:34
Speaker
person is thinking above the company, not companies above the person. so Not a team player, basically. Not a team player, right? Or telling these people something, another people something, transparency is not there, honesty is not there, integrity is not there.
01:42:52
Speaker
So it means they are not culturally aligned. And you can sense it. Some people will come complain to you, hey, what kind a manager this is, like is telling something about this person, about that. So, you know,
01:43:04
Speaker
What happened is not happening. Everybody has an interpretation. So everybody tries to interpret interpret their own version and share it with other person. But in reality, something else has happened. yeah So that kind of promotion, if you do, then you are not focused towards the result.
01:43:20
Speaker
You are trying to build your community and the vote inside your team rather focus on the result than the vote inside the company, you know. yeah Okay.
01:43:30
Speaker
Do you have a ah board of directors? Yes, we do. And these are all independent... What role does board play for an entrepreneur? You know, what's the role of the board?
01:43:41
Speaker
How does a board benefit an entrepreneur? It's like you... It's like a doctor and a patient, right? So the board is like a doctor, right?
01:43:52
Speaker
You go to the board, you tell that I have this disease and the board confirms, yes, I agree, this is the disease. Then board tells, okay, what's your plan? So you tell them that these are the medicine I'm going to take.
01:44:03
Speaker
This is my five weeks course I'm going to be on. So board is just going to listen to you. So this is a good doctor because they are all running a successful businesses. They all run billion dollar companies. They know how, so they're the doctors, right?
01:44:15
Speaker
And you go with your problem and you also go with your medicine because board doesn't have a time for the medicine. They tell you, okay, what's what's your problem and what's your medicine? Now you have accountability guy. Now every three months they're meeting with you.
01:44:28
Speaker
They have a last presentation what you gave them about your problem and then your roadmap. And this time you're going back to them. You refer the last presentation and tell them that this is how I gave you the problem. This is what we did. This is what my improvement plan looks like. The board will say, are you misleading me? You are just not giving the right answer. You didn't work on what you promised.
01:44:49
Speaker
So you are actually declaring your problem. and holding somebody else accountable to fix your problem you know observe your problem and sometimes you know just knowing that you have told the problem and told that i will work on it you have created an accountability chart there do you miss not having investors because investors bring in accountability they have genuine skin in the game, you know, in terms of making yeah your business scale up.
Financial Discipline and Strategic Growth
01:45:18
Speaker
Yes. Ten years back, you would have asked the question. I would be like more arrogant and say, absolutely not. You know, great business model, and investment. Today, if you ask me, I'll say absolutely right. We could have been a billion dollar company by now.
01:45:31
Speaker
We could have been a half billion dollar company by Right. And if I would have got a better financial investor, brought some more discipline inside the company, corrected me in my own way of demanding what the way I'm asking or running the business.
01:45:49
Speaker
There is a good. um science and engineering they bring in on board, which is very powerful. So because they know how to scale the business, they know how to build the numbers.
01:46:01
Speaker
um Only thing I don't like um about some of those investors, they don't know build the culture. So at least as long I want to, I'm a very great fan of culture. I want to have a good culture, trusting employees, trusting customers.
01:46:19
Speaker
That's why when customers come to our office, they love us, right? They want to come back to us every now. Our retention rate is almost 100%. We don't lose customers. Rarely we lose any customers.
01:46:30
Speaker
So it's all because of culture, right? um some other financial investors, if they don't see profit, they'll check it out. If we don't see profit, we can tolerate them for some time because we feel there is a value coming out from them.
01:46:42
Speaker
Right. So there is a difference there. What, ah you know, what was the realizations you had in the last decade? Like you said that you could have been a billion dollar business uh what were the things ah you know when you reflect what are those things where you took wrong calls uh you know we are all human and sometimes emotions can dominate you in your decision making right so when you have investors um your emotions are your financial engineering because they understand the number
01:47:13
Speaker
They have accountability to the investors. They have to return back to the market the money, what they have invested for. So with those financial engineering pressure, your engineering brain works differently.
01:47:24
Speaker
you know When you are entrepreneur, you are more connected with your people, connected with your customers, and you tend to work emotionally. you know And that's the recipe for... Like an emotional decision you took, which was not financially wise.
01:47:41
Speaker
so
01:47:44
Speaker
You know, um we acquired a company, right, with our own bootstrapping app approach, right? And it took a long time to integrate, took a long time to give us a result or return on investment.
01:47:57
Speaker
If it was so private equity guys, if they don't see the result in three months, six months, one year, boom, boom, boom, they will make changes. We didn't do that. We took a long time to see how the results will come in. Right. So and then if those kind of guys are sitting on the top, they have an art of looking in a deal. So they will immediately come in, tuck in another $50, $100 million dollars business with us.
01:48:22
Speaker
And within two years, we'll be $200 million dollars company. Okay. a okay But in our case, being a self-funded bootstrapping, that level of risk taking is betting everything. like you know And sometimes you don't take ba take a bet like that. But when people are financial engineering guys, they have been taking bets because they follow certain rules, principles of running the business, which is good.
01:48:47
Speaker
One has to learn the art of running the business. So i would suggest if... have to have any founders start the business i would say before you start spend five p guys or investors spend enough time with them and learn the art why they do the business become their student and then be the entrepreneur you will be better entrepreneur then just build your business and then later on in interact with them uh you would have missed the boat like i was young guy when i started nobody taught me all that right so i wish i would have learned all these things i would
01:49:23
Speaker
do a different kind of business today. ah Although, you know, I've interviewed lots and lots of founders, ah including founders who came from ah ah VC slash PE background. And I have not always seen That background giving them an unfair advantage in terms of being able to hit scale. It's been a mixed bag is what I have seen. so um But yeah, for sure, ah understanding that discipline often an investor is an extremely valuable skill for an entrepreneur to have.
01:49:59
Speaker
Yeah, nowadays the landscapes are changing. If you look at the matured market of digital engineering and AI engineering, there are a lot of players and a lot of PE guys, they have board members who are now investing and sitting as a partner who have run the company, who have founded the company. So they have actually brought in the founders on their side to go and scout for the companies and evaluate the company with whom they want to acquire, right?
01:50:26
Speaker
So, unlike just the financial background guys who only know how to book the books and how to raise the capital and do not know the strategy of the business or engineering of the business, now you need the operators on their side. So, the last 10 years, I've seen Many of the firms have brought in those operators with them and that has even different level of maturity. So even founders feel very comfortable to talk to them, interact with them in different language altogether. so
01:50:57
Speaker
Are you looking at bringing in PEs now for the next phase of growth? or Not yet, not yet. Right now, I think for a few years, we are set for our own growth very well.
01:51:07
Speaker
So I know it is a slow you know good growth. I won't say slow. I think we are beating all the this year. At least we will have one of the fastest growth here, which we haven't seen in the last four years.
01:51:18
Speaker
How much are you expecting to grow this year? I think it should touch about 35% to 40%. That should. you know Normally, industry this year, a lot of people are talking just sub-20% growth, where we are already hit all the numbers according to 35% this year for half a year.
01:51:37
Speaker
And this growth is on the back of Agent TKI, like Agent TKI deployment. I would contribute some to that. so So certainly Agent TKI will have a good contribution. And second, I will also give contribution to our loyal customers and the engineering team who are working with them. Customers are feeling more comfortable with our people.
01:52:01
Speaker
So more and more modernization, platform modernizations are coming to us because of the trust and the depth of the relationship. um you know If they have a choice to give two projects to different partners, I think that's coming to us because of your delivery capability, domain, their likeness towards you. So all that is playing hard as well as they consider us a thought leader, value partner. They see the knowledge and experience through platforms. So they feel that we are where could good people to work with.
01:52:33
Speaker
okay okay uh i wanted to ask you this question earlier i forgot um why did you open delivery centers in brazil brazil etc like why not just focus on india i'm sure uh cost-wise india would be more favorable than brazil uh Yeah, that's that's a good one.
Enterprise Sales and Client Engagement
01:52:56
Speaker
So, you know, I mentioned to you that we became a digital matured company and it started delivering more digital projects faster, say 10 years back, right?
01:53:09
Speaker
And when you are working on a digital engineering project, it requires a lot of customer interaction. So customer has to tell you the requirements, see some of the screens and debate with you.
01:53:21
Speaker
And it requires a time zone to collaborate that. right And when you are developing also, they want a constant feedback on your development process in the beginning phases. And if you look at the East Coast, we are in New York and we are in Pune and Mumbai.
01:53:36
Speaker
So the time zone difference, we hardly get two, three hours of overlap of communication. And that two, three overlap communication is not adequate enough to communicate with the client because clients are also busy. you know, morning time, they're busy. They may not be available for you.
01:53:52
Speaker
So if they are not, and then you need their time after 11 o'clock, India is now sleeping, right? So the old days, the model was different. You define the requirement, give it to Indian developers, and they will build it and give it to you. Then your communication was not required that much, you know?
01:54:10
Speaker
It was waterfall. Now it's agile. It's a digital project, you know? So more communication required. And then Brazil, Argentina gives us the same time zone. So developers...
01:54:21
Speaker
studio guys, project managers, architects are in the same time zone. So now we are not paying them the US rate, maybe 50% cheaper than US rate, or or maybe at least 70% of the US rate, but you are having the people in the same time zone to interact with.
01:54:40
Speaker
And then um So they come as a part of our U.S. delivery engine and then India delivery engine can collaborate with our Latin development team and get all the knowledge base and deliver it.
01:54:56
Speaker
How do you crack enterprise lot of businesses come through references. So one CIO or president moves from one location to another and they have worked with us for four five years. They have seen our work and they are at a top level at New Place and it's easy for us to enter through the reference based approach. So that's one angle.
01:55:22
Speaker
The second angle is ah go to the biggest customers and you have innovative products like Rise and AI Agent. And when you show those Fulcrum One earlier days, low-code platform, they feel that we are the smart value provider. They don't have their current provider providing that.
01:55:42
Speaker
And as a differentiating company, we are able to open the door, right? So you've got to open the door, but opening the door is not the enterprise sales. Once you open the door, you need to map the top level executives as well.
01:55:57
Speaker
And you have an ability to interact with them and continuously build the image of your company that you are not a value provider to only manager level or director level or procurement level.
01:56:09
Speaker
You can be a value provider to the CIO and not only CIO, the business heads, the presidents and the CEOs. And if you have an ability in your company to scale the conversation at all levels in the client side, now you're becoming an enterprise customer.
01:56:24
Speaker
It means now you have an ability to um chat with them. And those customers are very, very seasoned guys. If they meet with you, they will have five minutes conversation and they'll pretty much know you are bullshitting to them.
01:56:37
Speaker
or you're really understanding their domain, understanding the pinpoint, and you truly stand as a good partner, right? But if you have an ability to do that, i think you've got a recipe for enterprises. And then you need to scale that business model in your company because only few people cannot do that job. You need to have ah the whole engine, sales and client management engine. They have an ability to...
01:57:00
Speaker
ah talk at multiple levels and speak different language at different levels, then you can do enterprise sales because one size does not fit. How you build these skills or are these kind of people?
01:57:13
Speaker
Do they start as ah like people who are developers on a project and then they are identified to have good communication skills and then they go into more client-facing kind of a role? Is that how it happens or you hire people for communication skills and put them directly into client-facing roles or...
01:57:31
Speaker
okay but because we have been in the business for quite some time so we have done both we have nurtured good amount of people internally if you look at our ceo he used to be project manager and delivery manager right when he was hired and then he became client ah service manager then he became director client services vice president then he became a ceo of the company but his competency has been very good with the client interactions understanding what client wants building the relationship And today he's the CEO, right? So people have been homegrown like that. There are a lot of engagement directors we hire from some of the big companies you named. um
01:58:12
Speaker
There are so many employees in our company who have come from many big companies. They still work with us. um and they are the acquired skills. They have managed the art of dealing on behalf of those companies because they have been dealing with larger deal size.
01:58:26
Speaker
So they knew how to handle. So we have done both homegrown and acquired experienced people and then they collaborated together. So with the acquired skills, we brought in the market knowledge.
01:58:36
Speaker
With the homegrown team, we continuously build the confidence. So I think the cycle has gone both ways for us. And engagement manager is typically someone who's doing both service delivery and ah farming for new deals.
01:58:52
Speaker
It could be three skills. Three types of people can be engagement manager. One, um who comes from a delivery management background. In fact, I have found some guys who are a technical guy who wants to be also or client management guy, which is rare, but we do have some examples like Common is delivery managers becoming engagement managers.
01:59:11
Speaker
But then we had a sales guys also became a very strong engagement manager. So they were hunters earlier. Now they have become strong farmers because they had inclination to learn the domain, learn the technology, and they have also become a very good engagement managers and they have become directors, also engagement directors, right?
01:59:30
Speaker
So all three comes into picture. The percentage for technical becoming that is very less, but delivery background very high. And sometimes sales guys becoming it would engage in manager. So that that percentage is also high.
01:59:43
Speaker
ah An engagement manager would essentially talk to the clients, get their feedback, pass on feedback to the team and also ask clients that can you introduce me to people in like like your peers and people in other functions within the company as a way to get ah more business. like That's what an engagement manager would do typically.
02:00:05
Speaker
It's a simple metrics in our company culture because we have two metrics for that. ah Retain existing, existing business and add existing new business. So retaining existing, existing also isn't a KPI for a manager because they have to retain the revenue. It means they have to Make sure what has been promised is getting delivered. There is a risk mitigation plan. If somebody is leaving, there is a backup plan to fill up so that customer does not get pinched.
02:00:33
Speaker
So there is a good amount of effort effort required to even retain the revenue and maintain the quality and standard, right? That's why customers come to us because we give lot of importance to that in our KPI and measurement there.
02:00:46
Speaker
And then existing new. So you have an ability to plant ideas, debate with the customer, show them multiple paths. If customer feels that if I sit with your engagement manager, I get five ideas.
02:00:59
Speaker
I sit with some other engagement managers, they say the guy, I just run away because he doesn't talk value. you know um So I would rather have a knowledgeable engagement manager where I sit, I'm not able to think through, but the 10 ideas comes to me that why I'm doing it this way, I have other paths to do it, or um there are better ways to optimize my cost. And this guy is helping me optimize the cost.
02:01:23
Speaker
And because I optimize the cost, I'll give more business to him because I save the money. So I'll give more budget to him. So his job is to bring existing new revenue. So that's the KPI. And then we have a third KPI, which is a net-net new revenue, which is for the hunters.
02:01:38
Speaker
So hunters are just gone to get new logos. So that's a different bucket altogether. ah Okay, fascinating. um Okay, I have taken enough of your time. ah let i want to thank you for being so generous with your time. i truly enjoyed the conversation. Thank you so much for coming on the show, Rajesh.
02:01:56
Speaker
Now, Akshay, you know, you covered variety of topics from you went deep with the AI agentic journey to the PE side, to the entrepreneurial side. So I think you extracted a good amount of perspective from me. So I thoroughly enjoyed the conversation and I hope the audience who are on this journey can find some meaningful ah content from here.