Adil Ajmal

I still keep trying to figure out what I do. But recently, for the last six and a half years, I had the pleasure of being the CTO at Fandom, the world's largest entertainment and gaming fan platform. And before that, I was the CTO at a fintech company called LendingHome. Ran not just tech and product, but also ran two of their business units there. So that was an interesting experience. And I've done similar things before that. I've taken four companies through acquisitions. So I just happened to be at the right place at the right time, but to learn from a lot of folks. The last one that we sold was TenMarks Education that Amazon acquired. So that Amazon, for a few years before that, did a social networking company called Posterous, which Twitter acquired. And then I was at Intuit before that, got there through the acquisition of one of my other companies called Homestead Technologies. Got to take a fun, larger role at Intuit, because they were trying to go from being a desktop software company back in the day to being a SaaS company. And so my team is the one that created Intuit.com into their SaaS marketplace, and a bunch of other things. Sold a photo sharing company before that called Foresight to United Online, the guys that own NetZero, Classmates.com, and a bunch of other stuff. So it's been a very interesting journey. And you've got some enterprise stuff before that. So it's been a great journey of just learning new things throughout, and just having to work with great people.

That's literally the news of the day. I think Meta just announced an 8,000 people lay off, I think yesterday or early this morning. Again, similar things. And I think two years back, people would even pause to think about it. Right now, it's just happening. And I don't think you even give it a second thought, unfortunately, right now. Not easy for the folks that are being impacted, but this is literally not even news right now, unfortunately.

I am on the outside. So I get to be a little bit more cavalier as I give this answer. But look, I think impacting people is a very painful thing, regardless of, even if you're doing it for the absolute best reasons, it is not something to be taken lightly from that standpoint. Now, I think for certain companies, especially some of the larger ones, there had been a lot of overhiring in the last few years, especially since COVID. And these end up being ways to clean that up a little bit. In terms of just going and saying, hey, we're going to reduce our workforce significantly because we have AI, I think that's a little premature. That is the drum beat right now, because that's what's coming to you from your boards, from your investors, and so on around what efficiencies can you get from AI and how many people can you get rid of? I mean, that's the question that CTOs get asked, unfortunately, all the time. Literally, since the first version of ChatGPT came out, the question was always, OK, so how many fewer people do you need without actually understanding what the implications are for any company? The backlogs are much bigger than what you can go and produce anywhere. And so the first thing was to figure out what efficiencies can you actually get? How much of this, as you said, is ROI positive? How much of this will actually move the needle before jumping the gun and saying, we need to get rid of people and stuff? So it is a mixed bag. Now, to say that AI is not having an impact or that people are still figuring it out, that would also not be accurate. AI is absolutely having a pretty big impact. I think how it's having an impact has changed. So if I were to go back about two to three years, for companies that were more forward thinking around AI adoption, it was more about how much productivity can we get out of people. And based on that, you could either say, OK, we need fewer people, or we can do a lot more that we wanted to do, which we weren't able to do before.

Now, I think the focus is less on productivity. Now it's more about how can you create more value? How can you actually get the product to market faster and validate it, not just actually shipping something, but using AI to be able to validate it faster and then being able to re-enter it? And so one of the things that I would say has changed is that instead of people just coming in and trying to do prompt engineering initially, it's more about context engineering. And being able to create a lot more shippable products versus saying, we can create more lines of code through AI, or we can have 20% productivity, or whatever. So I think that story is now significantly changing as you're going from more generative AI to agentic AI, and how your workflows are changing, how you're actually incorporating your proprietary data into it, and so on and so on. Now, going back to your original question around, should we be firing people? That's a very subjective thing. And it's not a right or wrong thing, and it should absolutely not be taken lightly.

A little hard to say. I think it's still early stage, and it's going to evolve a lot more. Now, what I would say, where it's actually getting really good at is to be able to get something to market and test it. Now, where it still needs a lot of work and where you still need a lot more human involvement is scaling it, making sure that it's enterprise ready, making sure that the security actually works. And then also, creating something new is significantly easier than going in and transforming or adding to a system that's already pretty large and complex. The context windows don't work well there for AI. When you're creating something from scratch, it actually has pretty good context. When you go give it a system that has hundreds or thousands of people working on it for a long time, there's just too much context that gets missed in that particular point. And so breaking those tasks down still makes it really, really productive to use AI. But to ship a new product in that particular environment, that's integrating into the existing products, that's way more complicated versus launching new things. But getting something to the market is way more faster right now. That entire paradigm has changed. And anybody who's not going to keep up with it is going to have massive problems.

Take a legacy system, right? That is just much, much harder to be able to just say, all right, here's my agent. It'll go and just take care of the entire thing. Still not very productive in a large system.

This is where the ROI part comes in. This is where you have to be disciplined to be able to say, if we're actually going and building something or if we're investing in something, we have to have a very tangible outcome in mind. We have to be able to measure our way to that outcome. And we have to be able to measure the impact of that outcome and then have the discipline to say, no, this is not working. Or the learning that we got was useful enough that we can go for a phase two for this. Or that this actually has to be stopped now. Unfortunately, most organizations don't end up having the discipline and then end up going too far on one side of the pendulum. And then what that means is that when they try to correct that, then they end up going way further on the other side. And that's not a good recipe for success for most organizations.

Processes of the people, absolutely.

So a lot of the fundamentals around best practices haven't necessarily changed. So if you had solid good best practices around how to adopt frameworks, how to adopt processes, how to do change management, those fundamentals still exist. Most organizations are just not good at it. The ones that, I mean, honestly, I advise a lot of companies. I've gotten to see a lot of them from the inside. And there are a few companies that are absolutely awesome at it. But then most others are not. They're trying to play catch up. And as you said, a lot of it depends on the people who've instituted those processes. So those who have solid processes have a much better success rate at adopting AI and being able to stay ahead of the curve with the new tools that are coming up with the new frameworks. I mean, take compliance and security, for example, as it's coming in. You can create agents out of the wazoo at this particular point. But having the discipline to figure out what they're doing, what data governance is managing, what they have access to, what mistakes they're making and stuff, most organizations are not thinking about that. It's a very, very real problem right now. And there are new companies that are coming in just to solve that problem, because this is going to be a massive, massive issue as we go forward. But that's just literally one example. Then it's around access, not just to this data, but how are you actually giving access to your employees? If you go back and look at two or three years back, it was more about which licenses can we get and which licenses can we roll out. That's not adoption of AI. That's just procurement. Adoption of AI is more about how are you systemically changing your processes, your mindset. How are you measuring the ROI of that adoption and continuously moving to change that. Those are things that are hard for most organizations to do as much as they would like to. And so this still gets stuck in those early stages.

Exactly. I think if you think about it, the paradigm has changed enough that it's not about which AI tool, which company is getting, and so on, for the most part. It is about are you truly changing your processes one by one for the long haul, adopting them with AI. And that's where your win is going to be. Do I have the fanciest tool? It's going to be about am I leveraging it with the proprietary information that I have? Am I really leveraging it across the different disciplines where I have the right to win? Those are the things that are going to make you win in the long term, versus just saying, oh, cool, we will get a lot of new tools and we'll just have it across the board in the company.

Correct.

There's a big challenge. My son is in the freshman year doing a computer science degree. And I think the way their entire profession is going to change very significantly, as compared to what a computer science degree, what it meant 20, 30 years before when I did it. The way I look at it right now is, any engineer who's either coming out of school or who's in their early career, even later on as well, especially for those that are in their early careers, it's about how well can you leverage AI to not just be able to produce faster, but to actually have a better result from a product perspective. And so remember the thing that I was saying around, it's more context engineering versus prompt engineering at this particular point? That's the thing that you have to be focused on. And so I personally think that computer science degrees are going to have to change to an extent where they will have to teach people how to understand what the actual user impact is, how to actually figure out what the needs of an industry are, and then being able to apply technology to it. Now, there's still going to be certain fundamentals as you go into more specializations. Those will exist, but those will become fewer. Your regular average engineer who's going and doing product development, if they're just learning how to code and create a product based on a spec that they were given, that's going to be a dying breed over the next decade. And so that's the hard part. And for those that are more product-centric, those will still continue to win. Because look, you'll find all sorts of engineers in any large team. You're going to find some that are very product-centric. And then you're going to find others that are a lot more focused on the technology aspect of it, or that are, to be honest, just coding the requirements that they're given. That last part is going to be impacted the most. The winners are going to be the ones that are a lot more product and user-centric.

I would say a little bit of a liberal arts degree to understand the human aspect of it, a little bit of a business degree to understand how you're actually going to monetize it, how you're going to take it to market, how can you validate the product that you're building. Those are really, really useful things for an engineer. Because then when you know what problem you're solving, you can actually be most well positioned out of anybody else in the team to figure out how can technology solve that problem better. But if you don't have a fundamental grasp of what that problem is, what that outcome is that you're trying to achieve, and that's not the thing that is taught in a computer science degree. As long as you've got that, then you don't really have much to be worried about.

It's a combination, right? Yeah, I absolutely think the curriculum for these degrees is going to have to change. You're not going to spend maybe four semesters of this programming in C++ or Python or whichever language you want to pick. You still need to understand fundamentals, but you need to understand how to actually go and figure out the human aspect of it, the business aspect of it, and apply AI to be able to solve that.

I completely agree with that. By the way, this is a problem that existed even before AI. And so when you have engineers in other countries that are not familiar with the projects that they're building, they're not familiar with the end users, the way they think about that problem fundamentally is very different. That's where you've got the software houses that will basically just go and build the requirements that are given, whereas in most good tech companies, especially those that are here in the US, especially you take your engineers to the customer. And this is not something that I learned myself or that I invented myself. I got to learn this from great companies and great people around me. You have to take your users to figure out what the business processes are. My first day at a company called Homestead, as part of the orientation, I had to attend customer support calls in the first day to get closer to the customer. And so Intuit used to do the same thing. I tried to do that in other companies where I would take at TenMarks, I would take my engineers and have them sit in classrooms so that they could see how the product was being used by teachers and students. And that's where those aha moments come. Now when you have engineers that are in other countries that have no connection to the product or to the user, it is much, much harder. And so that's where they're going to be at a more for disadvantage, because they're not just going to be given those requirements so spelled out. People are going to depend on them to understand what the outcome is that they're trying to achieve.

Exactly, 100, exactly. I mean, even if you don't understand the product, just the customer support standards are so very different. The US is way better at it than Europe, and Europe is better at it than Asia. And so when you're trying to build these outcomes, it's not just about the product, it's about the entire user experience. And because many of these folks have not experienced that firsthand, it's hard for them to imagine that or to build for it. So that is a very, very real problem.

Oh, this is just such a big and evolving place right now. I don't think there's a single topic that I'd want to pick that would be done in five minutes.

We could literally pick any of these topics and spend the next hour deep diving into it. And what we'd find after an hour is that we still have so much more to talk about.

Those are absolutely the most fun conversations and I would say one of the other key things in that is building your cohort in a way that people are dealing with similar problems. Somebody who's dealing with something at a very large enterprise level is going to have a very different conversation or frame of reference, versus somebody who's trying to build a startup. I remember First Round used to be really, really good at it. Now a lot of people have replicated that, but I've been part of their CTO council for a long time. And so they would host, they started with CTO dinners, but now they do it for pretty much every function, but they do it based on the stage of the company. So that the conversations you're having with your peers, the eight or 10 people that have gathered are actually relevant to the life stage and the problems that you're dealing with, versus again, as I said, somebody who's dealing with the 10,000 people enterprise versus somebody who's dealing with a 20% company.

All the life stages literally have so much differences in them.