The GCC Pod Ep 2 Part 2: AI Works Best from Middle to Middle, Global Capability Centers in Pakistan

And I think, you know, I was reading this explainer as well recently from, from a top consulting firm and I think that framed it quite well, right? Like these are systems across some of these use cases that you talked a little bit about that, you know, the goal is not to just assist, but as you said, like be a digital human, right? Extend, genuinely extend operational capacity. So that leads me to, you know, the obvious question, which is that these models are getting better so fast. What do you think, in your opinion is slowing companies down with this?

So there are two parts to it. The first one is if you ever sold to an enterprise business, the way they procure systems and software is too ancient, right? So we deal with enterprises all the time. So the one of the things I asked myself is like, is there a way before we deploy AI systems in the workplaces, can we do anything to transform their procurement process? So one of the things is like enterprises are just slow to buy software, not to adapt. They have some, I've seen some of the enterprises have incredible chief AI officers and people like that, but somehow even when I talk to them, they are just, their hands are tied to their procurement officers and they don't know how to get this done. So that's the one thing, adaption is an issue. Buying, sending to enterprise and then going through that procurement process, that takes lots of time, but that's one thing. But on the other hand, the challenge that currently what I told you is like AI mostly works from middle to middle. There are parts of the AI that needs to be done to start at the, that needs to be done at the first mile and that needs to be done at the last mile to make sure that AI in the loop works. And that heavily depends on how structured data you have, right? So one of the major bottlenecks for deploying AI system in the organizations is or is what kind of data health do they have? I mean, at the end of the day, if you, if you do so most of the work that we have to do with these businesses is organizing their data, which is basically setting up, bringing that data into a shape where we can train, model and things like that. So these are the things that I think one, procurement flows, how to get through them, how to keep waiting for six minutes. And then as you are deploying, as you are in a position to deploy a system into some of the use cases, what you learn is that you started from ChatGPT4, now it's 5.2 and how to go about it. That's one thing. The second thing is like there are some serious data sanitation issues in the organizations. That takes a lot more time to make sense of AI. You know, 2026, I have this one vision for the entire Code Ninja team. Make AI work for the customers. And internally that's, I mean just, I've reduced this to this one, like make AI work. And if you really see what does it mean to make AI work, what it means is like make sense, make sure that the data traces you have makes sense of that.

And you know, obviously like with this stuff about procurement and enterprises moving slow and all of that is, you know, change management, coordination, red tape, but talking a little bit more about the data layer of it and you know, the importance of having that data, like obviously that would then also sort of bring about a change in how, you know, data and software teams are actually structured in the enterprise. So what does this do to that? Like, you know, how do you see that evolving?

So I started talking about this at least three years back that, that every software engineer needs to be an intelligence miner. What it meant was he has to understand how data flows works, he has to understand how process orchestration works. And he also needs to know the domain of the business. If the software engineer can't do all of this, one software engineer can't do all of this. Writing code is not something that is valuable. That's one thing. So we need to make sure we need to update the JD of a software engineer massively. And these three things are important. The question when I asked off engineer is like, what's your interaction with data? Have you ever sanitized any data? Have you built any data pipelines? And that's the basic question. The second thing is of course there is soft engineering questions, but. And the third is how do you understand domains you haven't worked on? So for example, if a software engine is working for a chemical manufacturing plant, how, how would we make sense of it?

Right?

All these things needs to be combined in this one neural network that we have, this biological neural network. Only then you can structure teams that can solve problems. Writing software is a commodity. No one needs it anymore. I mean I can build, I mean, you know, there's a platform like Replit, but, but of course you're going to need a lot of LLM. But then we have built our own platform. Hyper Hyper builds enterprise grade software in split seconds and lots of AI is involved in it. So why do I need, I mean why do I need software engineers? So software engines are required where I get all these three things in 1. Biological Neural Network where the weights of that guy or girl is adjusted in a way that is very good in acquiring new domains, is very good in understanding data. And they're going to build a process around it. So that's the biological, that's required from the biological neural networks. This neural network is not available in the digital form currently.

What you're saying, right. I think this feels to me like it changes entirely how and also where companies build software. So is that a fair observation to make?

Okay, so I think yeah, it's very fair. AI. I mean again, ruliology, back to the basic rule, think about business. When you think about your business, you never think, okay, this is the business and these are the people. Had it ever come across into your mind? Like, okay, let's say you are running this factory or you're running this logistics business and you think of like, let's outsource people to someone else, the core people, the core who understand business. You always retain them and it's always, you never outsource those people. So business for you is of course what you do for the customer and your people. That's, that's always been the business for you. Right? So as we established from the rheology, AI is also the digital neural networks. It's a digital beings. You can't outsource AI as if. You can't think of AI as if something, okay, that's something other businesses has to do it for you. You have to embed into your business. So it's insourcing, as we like to call it. It's not outsourcing, it's insourcing to bring, to bring AI into your business rather than outsourcing AI workflows. Because at the end of the day, when you're outsourcing AI workflows, it does not mean, it does not mean, I mean you are just outsourcing workflows. What you're saying is AI would be useless if not giving it a data. So it means. And data is useful if that data doesn't know the ins and outs of your business and if another company is basically the brain of your business, then what you are doing then. So for example, if you're working with a software business, let's say, I'm just talking about, if you're working with a software business and they are doing all the brain work for you. So essentially what it means is they know all the rules and they can build their own supply chain business then.

Right?

So really what you want is AI is not a thing you outsource AI thing you insource, which means, I mean we've been talking a lot about this, like build your own global capability centers, build your own teams rather than just build your own AI powerhouses, rather than working with third party businesses who build software and AI system for you is not going to work like this.

I like, I like that phrase insourcing and it's not me.

I mean actually I follow lots of GCC blogs. So there's a company called ansr, so they talked about this a lot.

Of course, of course. And you know, you mentioned gccs, you mentioned ansr. So then that in this entire story, in this entire conversation, where and also how do places like then Lahore in Pakistan or Riyadh or Chile fit into this entire story?

So when we started coding we wanted to land up at a point. We are helping some of the most important businesses in America in following the highest priority problems and creating opportunities for knowledge workers in the emerging markets. Both come together as one holistic outcome. That was a starting point. So that's, that's the mission. That why we want to solve some very interesting challenges, help America win against China in the race of AI. But on the other hand, I also want to bring emerging markets, Pakistan, Jihad and Chile along with our journey. So that's, that's, I mean I just wanted to clear it out. The mission of Code Ninja. So of course the emerging market, what we'll learn is like AI needs lots of engineers who have the exposure to the frontier tools, right? There's incredible amount investment and time is in these markets that companies are spending to give exposure to those engineers so they can build tools and solve some important challenges for the businesses. So I think when you talk about these markets, Chile, Riyadh, Lahore, I think in terms of setting up global capability centers, research centers, these are the ideal places to set up. And since we are doing this podcast, of course I want to tell that Koninja is doing a lot of work in this part and we want the businesses to work with us, buy into our mission and would love to solve some important challenges for them this year. 2026, our mission is making AI work for the customers and the core of our model is setting up global capability and research centers for these businesses where we make sense of their data. We have their own engineering teams where they are integrated and embedded into their business. So basically that's the go goal of coding.