Has an amazing article, which we’ll link to the Harvard business review and just how fascinating the use of AI and sort of strategy. Not talking about content, not talking about this, but like that kind of big thought process that all of us are, well, you know, at least we can say, yeah, it’s like, well, AI is kind of changing and adopting.
I’ve been fanboying, just looking at this journey, his journey today, and then just where he’s going with AI and how he’s helping companies look at it and play with it is something that I think will be fascinating for here in the audience. So look, without much further ado and rambling, I want to welcome Hamza Mudassir to the show.
Hamza, thank you so much for coming on. I am really excited to have you here today.
Hamza Mudassir: Thank you so much, A.J. It’s an absolute privilege to be here. And thank you so much for your kind words and wishes, really appreciate it. Happy to do a deep dive into everything entrepreneurial and artificial intelligence, and hopefully give some food for thought to your listeners.
A.J. Lawrence: You live outside Cambridge now, right?
Hamza Mudassir: I live in one of the villages on the suburbs of Cambridge in England. The university is 20, 25 minute drive so not too far off, but still in England.
A.J. Lawrence: Yes. I love Cambridge. I’ve been a few times and yeah. One, amazing pubs. I always love going to the ones where like, this is where DNA was discovered over just a few minor things have happened there over the years.
You have such a great thing because I want to kind of frame this journey, even though this is about entrepreneurialism, your entrepreneurial journey itself led to you actually being a professor and having your PhD. Can you kind of take us a little bit on your journey here as an entrepreneur, from kind of working in product and kind of moving your way up the chain a bit to where you are now?
Hamza Mudassir: Yeah, absolutely. Obviously very fortunate. I always start off with a preamble, which is a lot of people don’t talk about how much they sucked getting to this point. Believe me, I did and I still do. But you don’t talk about these are like those are kind of like things you go like, okay, fine. But really like, very, very fortunate.
There is definitely an element of luck that has played a part in it and I’m very grateful to all of the people who have actually helped me get here. Even the ones that I might not have recognized, just given the complexity of the world. So I’m originally from Pakistan, very similar to you, A.J. Both my parents were doctors and so they thought I wanted to be a doctor and so did I. Until exactly like you, around my A levels, I was just like, Oh man, I can’t study that much. Isn’t it too much for me? I decided to go into Computer Science, I learned some programming and then went into the world of business, predominantly as an employee in various sort of product and strategy roles.
And I was very fortunate. I got a scholarship at Cambridge University to do my MBA. So in 2012, with an 8-month old daughter, basically sold off everything and came to Cambridge. And I was very fortunate. I did quite well there. It was one of the best years of our collective lives as a family.
And then got into Amazon in Luxembourg, worked with some fantastic people there. And then one of the places I had consulted with when I was doing my MBA at Cambridge. Those guys were a part of, at that time, a small video game studio called Jagex. It’s now huge. I came back for my graduation and they said to me that you need to come back to the world of video games.
And I play a lot of video games, still do. What I did not know was how challenging things were at that time at Jagex. And so I spent two and a half years doing a turnaround and I was very, very fortunate again that there were not that many people in the company at that point in time who sort of had the business skills.
It was primarily a very creative house. So I was the only MBA at that time in the company and I learned how to run a business. And effectively we sold it for more than $400 million two and a half years later. And I stayed put for another year and a half. I helped the Chinese investors list the company on the Shanghai Stock Exchange.
And then those four years, I felt like my mission was done and I got headhunted to go and work in an advertising technology company called Adstream. Very similar challenges to Jagex, but this time around I knew what I was getting myself into. I had a partnership with the CEO. He’s a great guy. The board was really good.
So I spent another two years. It was their turnaround, brought the company back to growth and profitability, and then sold it to American Private Equity for hundreds of millions of dollars. And that was all fantastic. In between I was a non-executive director and an early investor in one of my friends’ AI companies out of Cambridge and it was focused on education. And yeah, we ended up selling that as well around the same time.
So in roughly five years, I ended up selling three companies. Again, as I said, it’s very fortunate because it usually does take longer. Around that time, I felt like I wanted to go back to my roots, which were sort of building products, making new stuff, which was more interesting for me. And so I ended up starting my own consulting firm. This was around six years ago. A couple of years later, because of all of my misadventures in my corporate life, Cambridge University gave me an honorary fellowship in Strategy. A year later, they asked me to teach.
So I now teach at the business school at the Master’s level. And I teach quite a few things. I teach like sort of classic strategy, which is mergers and acquisitions and build, borrow, buy choices that companies can make. But over time, just given my technology background, I have ended up teaching an elective there, which is quite popular.
Effectively, it’s a strategy in the age of artificial intelligence. And I also teach things like ecosystem development and multi-sided platforms and a bunch of other stuff that comes with it. I don’t want to bore too many people with too many technical things. But yeah, these are interesting classes.
I’ve managed to remain employed for the past four years, but feel like an imposter all the time. And yeah, Strategize came out of the university in partnership with two other professors. One of them is the Pro-Vice-Chancellor of Cambridge University, and the other is the Chair of the Strategy department at the business school.
And the three of us were, and the company started off incidentally as an education technology company. And it started off in 2021, the three professors got together and they were like, man, it’s really hard to compete with PlayStation, Xbox, Twitter, and TikTok in the class. We are genuinely not worried about other professors.
There are great professors all over the world, right? So it’s not something that keeps us up at night. What basically was quite jarring for us was the fact that you’re teaching and the students are sitting on TikTok or they’re playing a video game or something in the class. And so we thought, you know what? We need to compete with these platforms. We shouldn’t be competing with each other.
And so we ended up making a digital twin of the US automotive industry, mapped at a data level the path that Tesla took to go from zero to an absolute juggernaut in under a decade and a half. And then we made a video game out of it and now we teach. So it’s quite popular.
It’s called the electric car revolution. And you basically get into the shoes of somebody like Elon Musk. You can choose obviously other kinds of avatars as well. And you make a bunch of decisions. You compete against AI in a hyper realistic model of what happened. But the cool thing is that it is a procedural model, i.e. there are no rigid rules in it. So depending on how you play and the AI plays, you can just completely change the face of the industry. So it does not necessarily have to get electrified in a year and a half.
So we ended doing that and Strategize.inc board.
A.J. Lawrence: No, I mean, it’s so funny because when I got my MBA way, way back in the early 90s, we had competitions and stuff. But like what it reminded me of was of all the books I read, even though I didn’t really get into operations, I was too busy trying to chase girls, but it was the goal.
But there’s the famous book on operations called The Goal, which is just sort of like basic intro, you know, primer into operational structure. But in a novel sort of like the problems, yeah. I’ve seen it. And I remember that sticking out because it was so straightforward to take in.
And now with all the extra stimuli, for lack of better term, having a game and being able to learn and play with the lessons stuff I learned when I was in MBA, because I did go kind of straight through, I had a professor who liked me and kind of was like, you know, there’s a fellowship that no one’s ever applied for. It’s back before everything was online.
It’s filed in the role in the scholarship section so no one has ever seen it. I think it would be hilarious if you do it no one’s ever applied. It was like so, yeah, I had an accounting fellowship. But still.
Hamza Mudassir: And you’re a CMO.
A.J. Lawrence: It was just because my professor, yeah, it was like, he just thought it was funny that I would apply. But to the point is there was stuff I learned and heard that wasn’t relevant until years into my business journey. I had started businesses for companies and none of it made sense until I kind of got past that rubbing stick phase in my own.
So you creating this game, I think is really fascinating because it helps to bring stuff to where, one, the students, and then, you know, and then let’s kind of talk about how you’re bringing this now to companies. We see a lot, there’s all this like, you should be doing X. Great. There’s so much. Like every business book in the planet is pretty much the same, just with a different flavor to it.
You know, it’s like they change the vocabulary. They change the pacing. Instead of five phases, there’s 20 phases, there’s two phases, or there’s trademark XYZ strategy, whatever, which is the same as something else, which is the same as something else. And you go back, how we bring it into bear, I think is what’s interesting.
Because it’s like, you can always be told something to do. How we as humans bring that in kind of impacts our journey as entrepreneurs, the luck surface we’ve created, how well we bring it? Where you’ve started, and I’m going to jump now to Strategize a bit and where the article kind of started leaning towards was like, there’s more looking at AI and looking at it.
So first there’s like this ability to better understand how strategy actually impacts your own journey by playing in it and seeing the trade offs immediately. But there’s also this ability to have a lot of this knowledge brought to you in a supporting role. Those are my words. These are not what you kind of put out there, but that was kind of what I was reading between the lines of like, wow, there’s this ability to kind of have, you know looking at how well AI did in the competition, but of course, doing.
I know I’m on some boards and I’m like, yeah, I probably would fire the AI too. But you know, looking at what was possible as you were concluding in that of like creating that support structure, combining twinning CEO with it made me think about all this_ we should_ that is so hard in real life to consistently bring to bear as an entrepreneur. It’s like, I know I should do this, but I also have by fires. I also have all this. I have a sick child at home. I have to do this. My wife wants to make sure I spend more time here. You know, I haven’t seen my friends.
It’s hard. There’s a lot of noise. Where I kind of see you is on different levels from small to large companies is this ability to reduce the choice to a higher opportunity set. So instead of like, well, I have 50 things I can do, 10 of them are probably going to be really helpful and outperform other choices. You know, that strategy kind of like where’s my optimal thing? AI can help you sift through that with the right data. We’ve talked about making a little garbage in garbage out.
But that’s where I was just like, Oh my God, you’re doing what I’ve been wanting. Tell me like where Strategize is, where you’re going now from this, that you built this game. Sorry. I stepped all over you. Where from the game to Strategize and now to the research tests that you guys did and wrote up in Harvard to now working forward with companies?
Hamza Mudassir: Number one, I would say that sort of you summarizing why something like that is needed in the world today is absolutely spot on, right? There’s just so much noise. There’s just so much optionality. Even as an entrepreneur, you know that even if you fail in so many words that you can still bounce back. You can still get a job, you can go and do something else, right?
There’s just so many answers to relatively simplistic questions and the noise is so much. It’s not just about what do I do with my company but the fact of the matter is that your company is very much linked to your life and then where you live and how you live. And so it’s just, there’s a so much noise in general.
And so you do need like a safe space to reflect. But which brings us to the second point is who do you reflect with, right? I mean, you could make an echo chamber of your own brain, but with most CEOs, even in my own experience as somebody who was in the C-level roles within relatively large organizations, these tend to be extremely lonely positions.
And as an entrepreneur, and you know it, A.J., and I’m sure your listeners know it as well. It is very, very lonely. Even if you have partners in the business, even then it is very lonely. And the reason for that is that when you sort of take on that entrepreneurship role or a CEO role, it’s really hard to have an equal that you could safely reflect with.
So even if there are people in your your own team or there are minority partners or there’s a board or whatever, the chances are they all have their own agendas. Even your mentors would have their own agendas. Like the more I grew up in my career, the more I found that my mentors started getting really shady and more fixated on how to make more money with me than to actually guide me.
And so I found that to be a really difficult problem to solve. Where do you find someone who understands you well enough but does not have an agenda? And how do you reflect in a safe space? And how do you know what you’re reflecting on is objectively a good idea or a bad idea? That’s the part where I think naturally it’s the burden of leadership that you need to go through.
So how do you solve for that? And that’s the part where our research sort of indicated a couple of things, which we have built large parts of already and deployed it with sort of a variety of customers. But effectively what it tells you is this, that you could have an AI that can effectively not just navigate a digital twin of your company and make great decisions and do it in a variety of scenarios that you can learn from.
But over time, if you interact with it enough, you are effectively creating a digital twin of your own self. So as CEO, if you engage with our product and you know we have a strategy sandbox, the first few times it won’t know you that well. But as you engage with it more and you sort of make your choices, you have conversations with the AI itself, you’re basically creating a safe version of yourself with digital memories of you in that sandbox. So everything else you saw, which was basically can, uh, the questions about can AI, large language models, can they make good strategy decisions? The research says absolutely. Does it need a lot of high quality data to do that? Yes, absolutely. And those are things that we have solved and actually productized it. We were able to do it as part of the research.
We just did it because we thought somebody should, right? But the thing that we are really excited about in which is what we are building is that if you play in the strategy sandbox for enough time, and by enough time, I don’t mean three years.
I mean literally maybe a month or two. Whatever the sort of engagements you do, you have a digital twin of yourself as CEO. And that is incredibly liberating because that same digital twin can take decisions sort of on your behalf in a virtual setting and for you to see the impact of those. So that’s one.
But the second thing is this, that large language models are incredible in their breadth. So for example, GPT o1 is being trained on one trillion lines of information. One trillion lines is all of humanity’s wisdom being trained into us, basically a machine’s instance. And that is an incredibly powerful thing to have if you know how to harness it, right?
So for example, in our game or in our research or in our products, we harness that unique knowledge of human intelligence when it comes to business strategy and using that part of that knowledge and using that to sort of strategize about a company’s future. But there’s just so much else that large language models can offer, like from Greek philosophy to the description of the nature of black holes, like you name it and this thing can cover the breadth is almost infinite.
And having a digital twin with that breadth means that over time, your conversations will get deeper, but the answers you will get will be very creative as well. And there might be angles because as we grow older and as we grow more senior and in a lot of cases, a lot more successful, we get myopic. That’s just the nature of what us humans are, right?
And just having a digital twin with that breadth of experiences that it has learned from human history and being able to sort of think creatively, we can, for example, in our products, we can control how creative the answers are and to a very precise degree. So if you want really creative solutions to some sort of a challenge you’re facing within your business, you can get very creative answers that think very literally. Even the smartest of people, you can’t do it.
A.J. Lawrence: And you know, it is really, as this is going, cause you were talking about its ability to kind of give you insight building it like just recently. And I know this has hit the store. This has kind of hit all over the internet and all this because ChatGPT has memory and you can add things to it.
Similar to what you’re talking about, some of the prompting that’s coming around and people are sharing of like, can you tell me what you know of me? Seen things that have gone much deeper. What I’ve done, especially when I’m working on some analysis of something, I will ask it based on what you know of me and what I’m trying to do here.
Structurally criticize me and then what am I missing? Where are my black holes here? What have I just completely missed? And it is fascinating because it’s stuff I’ve known I do. I rushed to the, you know, I’m a doer. It’s also my weakness. I am someone who’s able to get things moving. I have this great ability to convince people, Hey, let’s go do this.
It’s that completion, working through the details to the execution. And I’m like, holy, it’s like, I’ve worked with business coaches and it takes months before I can get this type of detailed analysis. Or working with visors in my companies I’ve had in the past, you know, and this is just a few months of playing around without even realizing it was there.
Okay. So we’re talking about the generalized this that, where is Strategize working now? Let’s talk about that. What are the client types that Strategize works with? Cause I know you have these great games, the auto industry one I was playing around fun with. But like, who do you work with?
Hamza Mudassir: That’s a really good question. So we have very early stage, right? So we are a bunch of nerds who got together and wanted to just build stuff that either solved what we felt were big problems that somebody needed to do, or just because we felt it was cool. The reason we ended up in Harvard Business Review was that in April, I think I saw a paper by professors from Stanford and Northwestern and Georgia Tech who had used large language models to simulate a decision making by state actors when it came to nuclear warfare.
And it turned out that 6 out of 10 times the world would go into nuclear warfare. And so that was very interesting. And you were like, okay. And that was a deeply charitable paper in which the assumption was that never go nuclear. It’s a bad idea, which I completely agree with, by the way. But it did not take into account any human sort of tendencies under the same situation.
And the fact of the matter is that we’ve had nuclear testing. We’ve had nuclear bombings in the past. While we can hold that absolute standard towards an AI and say, bad AI, it’s 6 out of 10 times. You should not be blowing people up. But what would humans do in the same situation, right?
Humans do press the button every now and then. So we had these users, we had these students not just from universities like Cambridge and South Asia and so on, so forth, playing the game. We also had executives from large banks, from all sorts of other industries playing the game to just sort of sharpen their strategic skills.
Because it’s a very easy game to get into, but it runs really deep and it comes with a lot of theoretical lectures from Cambridge professors as well. It’s just like as a package, it’s a nice like a booster shot for strategy in case you need it.
So we were like, okay, so we have all of this information about how humans would make decisions in similar circumstances. So can we get the AI to do it? And we deep linked GPT-4o. It was not an easy sort of engineering task. But when we did it, it came back with some incredible results.
Like there is a drawback very similar to the drawback that I mentioned in the nuclear study from Stanford and Northwestern is that the risk propensity or the ability to take risk was actually very high despite telling the large language model that this is real life. It has a tendency of taking higher risks.
And so we saw the same thing and that sort of high risk, high return was a problem there. We’ve solved that problem, by the way, now that we can control the level of risk or the risk appetite of a large language model in certain ways, and we can control it quite well.
When we did that and the difference in sort of returns on sort of key performance indicators like sort of virtual market cap or market share or profitability. So very similar situations. The difference in results, if I park aside high risk tolerance of large language models, the difference is not a small 20% difference or a 30% difference. The difference is like 900%, 1200% difference between AI’s ability to generate profits in the same situation versus the very best human beings in the same situation.
The value gain is huge. And yes, there are problems that either we have solved or are we going to solve. And so we saw that and we felt that that was a big opportunity. But we are a very small company so we couldn’t really get industrialized to that rate.
We’ll probably get to the point that maybe late next year where our products will, it will be easy for us to spin up sort of these what we call as dojos, these strategy sandboxes, within a matter of hours versus a matter of sort of weeks because we have to fine tune them.
So we’ve worked with the UK government to sort of strategize on the impact of cybersecurity breaches and cyber warfare on a variety of different UK industries. We effectively simulated the entire of the UK economy. And we did that within seconds. And that was really cool. We’ve worked with a lot of, obviously, universities from an education perspective where we started off. But ever since the Harvard article, we’ve been approached by quite a few different kinds of companies. You know, we’re sort of building those capabilities now for them.
When it comes to what is our ideal business, I think that technology is so sound that we think that we don’t want to actually just focus on banking, for example. Actually, no, we don’t want to. We would want to cover, we could do commodities training, we could do video games. We could do quantum computing. We can do just about strategy for just about any kind of industry. The best way of imagining us is we are a SaaS business, but that S, the starting S is actually strategy as a service. So imagine you go to a McKinsey and they’ll charge you a couple of million dollars for giving you a bunch of inputs and ideas.
Some of them would be fantastic, but it would quickly get obsolete, right? And in year’s time, sort of the data you do working with. Any updates on that strategy means you need to go back and pay McKinsey another million dollars. We did just take that away. We have basically used artificial intelligence to allow you to access strategy, not just which is two years old based on a bunch of random assumptions, which is based on data which is as old as one day old. And it remembers the strategies that you liked. It looks at what you’re doing and it can update it accordingly.
For example, you have your classic segmentation, you’re a marketer. You’ve been a marketer for some time. You’ve had classic segments, these are budget conscious users, these are elite users or power users or whatever. If you were to sort of take that view and try to explain how electric cars became as popular as they have, it would be really difficult to predict.
There was this really funny thing. I used to work for a telecom company in Pakistan and McKinsey, bless them, they came over. This was the very early days of deregulation. They said that given Pakistan’s sort of economic environment, the fact that everything is really, really expensive, we don’t think there’ll be more than a 100,000 subscribers in totality for mobile phones in Pakistan. And this was like 20 years ago.
And they use the same sort of segment thought process. What they did not account for was how rapidly the costs would go down, how brutal the competition was going to be, and how cheap things will get and how quickly. And so that rigid sort of segmentation model told them 100,000. I’m very pleased to share with you that today in Pakistan, there are 110 million people every month who use mobile internet.
They got it wrong, right? And the thing is that because they went for the highest probable, easiest to accept. Strategized solutions provide you strategy, which is not just, we can say, well, given everything we know today, this is the likely outcome.
But what if there are sequences of unlikely outcomes that are happening? Very similar to what has happened with AI before the transformer model, which was a massive, which is what ChatGPT is. It was discovered by scientists in Google.
AI was basically on a bit of a stalemate. Like it was expensive. It could really look at only very small, relatively small data sets compared to what you can now do with large models. But there was always a 5, 10 percent chance that one of these years there’s going to be a breakout. There’s going to be a breakthrough. And that when that breakthrough happened, the entire cost structure, the industry shape has completely changed.
And so segments from four years ago, A.J., as you would know, like five years ago, aren’t as relevant today in today’s market. And so the way we build our things is it is based on very complex chains of cause and effect, but they are very, very discrete. These are basically, they account for low probability wins and losses.
And such unlikely situations that if they were to happen could change things completely, could disrupt things completely. So we’ve worked with the UK government. We’ve worked with a bunch of sort of these very new emerging markets, but we’re also in, for example, in discussions with companies that do fraud detection and credit card transactions.
And so how do you predict when the next big breakthrough would happen in fraud? Not in terms of protection against fraud, but in terms of fraud itself. And what do you do with AI with that? And so we work with a range of different challenging cases because the more varied cases our technology tries, the better it gets.
We are in that stage for over the next one year where we would like to work with certain companies in certain sectors and we don’t want money to be the reason why we can’t. It does not mean that we can work completely for free, but it means that we will always find a way to work with.
If it’s an interesting enough problem, we’ll be obsessed enough to solve it. And we’ll find a way to work with a potential client around that. So that’s one of the reasons that I actually wanted to come on your podcast is simply because your audience is full of interesting people solving a lot of interesting problems. And if you could partner with even a couple of them, it would benefit them and we get to learn in the process.
A.J. Lawrence: Yeah, as I’ve been listening, it’s really fascinating. Like a lot of things that stuck to me as when I was younger was just sort of Moore’s law, the idea that it was like that made so much sense when I was in my 20s cause I remember like my dad making fun of me for playing for computers when I was 10 years old. Only people with broken glasses sit in the back room. Now, if I had listened to him, I would’ve had a very different journey.
When you look at it and then you kind of take like, okay, understanding Moore’s law, you can get how it’s implied. There’s unintended consequences. Okay. Well, where does that match? How about solar power capabilities, energy, battery capability is all of a sudden it’s growth capability is changing. It’s not rocket shipping yet, but the ability of batteries is decreased. The cost structure and the amount that can be stored is increasing and decreasing at the same time.
How does that impact it? We’re looking at this rise of the super need of electronic. Everyone has this like doom and gloom, but it’s like, well, we’re getting better. Our ability to absorb electricity, absorb energy out of sunlight is increasing. Oh, how does that impact?
If you have these large language models that can understand multiple patterns and interact where human, we could probably handle. We know they exist and we build mental models and we try, and that’s, if you’re really good, you can balance a few in your head.
But the large language model can give you this pattern, this pattern. Like, what are the known unknowns, unintended consequences? Well, we can’t do it, but we can know what has happened under anthropological returns when changes in technology occur and then model out how does that, and then taking that to an entrepreneur.
It’s like, okay, I’m doing this. Cause all right, just a quick example. I ran for a friend who was selling his company. They were stuck in some back and forth and he was like, all right, we’re really not negotiating. We’re just kind of talking to each other. So I use Never Split the Difference as a negotiating style.
I took negotiations more than 25 years ago when I was in biz school. It was very good in advance, but like Never Split the Difference is interesting because it comes from the FBI’s hostage, you know, release hostage, whatever program. And so it came a little bit differently.
So I said, all right, use this as a framework, here’s the transcript of all the conversations. Here’s all the emails. Let’s redo this. But after a while, what it said was, Oh, you’re missing. This is an interesting model. This is a good model, but you’re not evaluating it also by blah, blah, blah, strategy, structure of unequal power and all of this. And it kept suggesting ways of mixing some models around how to do this.
Long story short, we were able to increase the price by another 150K out of it after they had gotten deadlocked. So I was like, well, you know, for a week’s worth of playing around. Yeah. I was like, that was fun. I didn’t get anything out of it, but it was still really cool to play with that. And to see sort of, as we kept adding stuff, o1 preview kept saying, well, it’s nice, well- sorry, I was doing it with multiple and I was feeding different thing, trying to daisy chain because you can’t upload stuff.
Hamza Mudassir: So you were wargaming. Very cool, very cool.
A.J. Lawrence: It’s like, all right, you can load stuff to o1, but I could create a seed with Claude and then I could put the text into. I need to just build an API and have my own thing. I just, I don’t know. I’m lazy sometimes. And I just kind of have four things up and I kind of copy and paste between them.
But it was interesting that it was suggesting we were too locked into our strategy, and that I found really interesting into it. And then giving it, okay, if you’re finding that, what are our blind spots? What should we bring into bear? Then we kind of got some breakouts and some open discussions that it was like, okay, we still for lack, you know. And I did at one point say, give me a breakdown of what you’re, it was like 75% never split the difference and forgetting the other thing, and then a little bit by the Harvard, whatever their core strategy zero to one or whatever the stupid, only a complete blank on what their strategy negotiation pieces, whatever the book that came out of that.
But it was just really like, it was clunk work and it wasn’t real time. And it was, okay. You understood that it was both like, there was someone who needed to actually understand some of this. And then a junior role of copying and pasting and following up and making sure there wasn’t garbage coming. That’s interesting how that’s going.
And what you’re talking about seems very much like that. It’s like, okay, here’s what’s possible. You kind of start pushing your direction, we’ll play, we could cut back, we’ll give you that kind of, oh, this is interesting, but then help you broaden.
There’s a really interesting, and I wish I didn’t know, I would think you had talked about McKinsey. Someone has modeled out the whole McKinsey research and their-
Hamza Mudassir: Really? Amazing.
A.J. Lawrence: presentation that like 100 page whatever has modeled it out into a series of prompt structures. And I’ll add this to the thing and I’ll send it to you as soon as I dig it up.
Having read some McKinsey, I didn’t work there, my wife did. But having read some decks and stuff, it’s like, oh this is like reading 1.5 pass before client gets it, before it’s cleaned up and all this. But it’s like, okay, would I pay 200,000 whatever they charge for that thing? No. But this was done in 45 minutes. I mean, it was really just based off of a case study, not even like feeding it data.
Hamza Mudassir: Yeah.
A.J. Lawrence: So it was like, so he showed an example of a case and then, Hey, replicate it. I just modified the case study based on an acquisition. I’m looking at acquiring a company. And then I said, okay, my concern is, and this is like a market research firm. And like, all right, market research in the time of AI, is this a business anymore? How do you change into the stuff you knew but deeper and all the things. Then I had it play McKinsey’s versus Baines versus Boston Consulting, their strategy stack styles. And I had it come. It was like, so yeah, this is what I saw you doing.
So, okay. You’re working with these companies, you’re going to start experimenting more. When you talk about the type of problems, because you see this as being not an industry, but as sort of a condition, in a sense, figuring out who’s the person you most want sitting, playing with your dojo.
Hamza Mudassir: The C-suite. We’ve had interest coming in from, I mean, basically people who can make decisions. I think anything shorter than that, you won’t get the best out of it. I just explain how this thing works and just so we understand just the scale and scope of what it can do. So the first thing is this, that with any large language model, you are basically tapping into decades, if not centuries, of documented human wisdom, right? Across a variety of different disciplines.
And the way it does it is that it is a pattern matching machine, but the pattern matching machine has learned the patterns out of basically most of human history. And so it can go down a rabbit hole, which it knows, where it thinks the answer is precisely there. And that’s one of the reasons that it was able to do as well as it did.
So one of the things we do is that we use something known as a hierarchical autonomous agent swarm. So we don’t use one. Basically you said, look, we’ve put together two, three sort of competing windows of GPTs and we copy paste things between them.
And of course, as you said, it’s a manual process. It’s also very taxing for the person who’s doing it. There’s only so much you can do. We basically spin up hundreds of large language model agents, depending on the situation. And so whatever the situation is, you can go into a dashboard and really tweak each of those feed settings if you want to, or you can just maybe give it four lines about the assumptions you want to test.
And we basically, it spins up these large language models that then autonomously take on different roles in a scenario. In that scenario, then they play as rational actors based on the personalities and the conditions you give it to them. Each of them will take on a role. It could be a competitor, it could be a lobbyist, a policymaker, somebody in the government.
There’s also the protagonist, which is you in that company. And so it will spin it up, play out what could effectively is a very long game of dominoes. And it basically just plays on and you give it a time or a goal and it’ll come back and tell you what exactly happened. That happens when you give it a certain scenario and it goes back to trillions of lines of information and then tapping into human wisdom and the role you’ve given it, it’ll come back with certain decisions and probabilities.
But we don’t stop there because our scenario, there is only a 0.000001% chance that that scenario is going to happen. So what we do is that we then spin up a million scenarios.
All of them are a little different from the other one. And we get the large language models to play them over and over and over and over again. And what you then effectively get is a view of a million possibilities. And what that does for you are 10 million possibilities, depending on how wide you want to go and how creative you want to be.
And out of that, it can come back and give you 100 routes to achieve a certain goal. And that is the crux of what this does. So we are not writing brand new technology per se. We are going back and looking at the history of human intellect, culture, whatever, across a wide range. And then we create millions and millions and millions of likely scenarios that could happen and get the large language model to deal with it then. And from there you can pick out, and that’s the part which is the last thing. We believe, and right now it’s a belief, but hopefully in a couple of years it will become a fact, that the age of frameworks in strategy is coming to an end.
So like things like five forces model, it has not aged well. Like I can tell you, we basically teach it because people need to know about it. But I don’t labor on that framework for too long because it does very little explanatory power of causality. And the reason was that, that these frameworks were invented by the BCGs, McKinsey’s, academics, whatever. At that time, there was not enough data and you wanted to simplify a complex world because the human mind can only keep so much information in one go.
But what if you did not need to do that? You don’t need to take any shortcuts. There’s so much data out there in the world today and even if there’s data which is missing, you can upscale it. You can do a really good projection of what it could be and then you play out every game that can be played out.
Then why do you need a framework? You don’t. You just need to see how things would play out in certain situations. And yes, some part of the wisdom of the large language model is going to come from sort of the history of frameworks themselves. But that’s not the point. The point is you don’t need those frameworks anymore because you don’t need 10,000 analysts to dream up a million scenarios when it can be done in a couple of hours.
And that is the goal we are moving towards. And you get to choose what is the strategy that works best for you to help with whatever framework tells you whatever. How do you explain, for example, Elon Musk’s success? He does not follow any strategy rule books.
It’s really hard to get that. And that’s sort of that. And the more disrupted the world is getting, and the more data intensive it is getting, the more useless these frameworks are becoming. And I can say that as a business school teacher.
A.J. Lawrence: It’s so funny because, yeah, I mean, it is that like idea, like for a long time, I always thought my ability was to sort of, the value I brought early on was like, Oh, I know how to bring context from multiple viewpoints, things inside and out.
I could take something interesting, you know, steal and then make your own from different things. But looking at AI, it’s sort of like, Oh, this does it a thousand times, not even a thousand, a million times better than my ability to do this. And it’s ability and then it’s ability. Wow.
All right. We’re kind of going to the theoretical. We’ve talked about, okay, if this is of interest and you think you have an interesting thing, consider this, reach out to me or Maz directly about this. I think the idea around what we’re talking about as being applicable to people who are listening now is you need to kind of be looking at these frameworks.
Yes, there’s some jury rigging going on, but I think more so is looking to make sure you have the right data in place to then start looking at decisions and start seeing what tools are out there. Follow things like Strategize, because this is going to get these co-pilots eventually, even beyond are going to be rapidly coming.
Agents are coming through seeing stuff. Tableau is really interesting where they’re moving from that sort of dashboard, here’s your insight, thinking to role of agents. And I just don’t like the company Tableau, having back in the day done work with them, but I get it.
I’m always hoping they’re a little too early, sort of like, all right, they’re too far out the curve, but probably not. But it’s like, how do you start thinking about communicating your need? Start thinking about this as you’re listening and look at what Hamza and Strategize are doing.
You could create smaller capabilities by, like I said, cutting and pasting. You can get a little bit more further and play with APIs and kind of create your own black box, do things there. But it’s like this is where we as entrepreneurs need to start playing around with that and diving in.
As you look at this and yes, you have a couple in the next year or so where you’re trying to create this great virtual CXO or let’s just call it, AI CXO, to kind of partner and all the fun. There’s so many fun things we can go off with, but like, where do you see this taking you on your own entrepreneurial journey? What’s that going to look like as here you are, you’ve had some great exits. You’ve worked in some great companies. You’ve developed this capabilities. You’re doing some really cool consulting. You’ve become this professor. You created this great game, the article, you’re starting to be thinking out at the edge of where this industry is going. Where do you see your own entrepreneurial journey going as you bring Strategize out to the market?
Hamza Mudassir: Such a good question. I think that for me, Strategize is more than just about making money and not, saying that again, not a charity. It’s always good to make money. But I think that for me, Strategize is trying to solve a lot of stuff that for some reason a large amount of people have not found the incentive to do so.
Partially because of maybe technological, monetary reasons, or just time. People don’t have the time for things. From my perspective, ideally, I would say the Strategize’s goal over the next five years is to basically give you a supercharged McKinsey to anybody and everybody who wants to make money through business.
So it’s not like they are a doctor or an engineer or something, but they want to start the business. I think in five years time, maybe six years time, we want this to be like a co-pilot that you don’t want to put down. And if it is something that provides you outsized value way more than you would pay for a subscription.
It is not just something that you will help you do X, Y, or Z or does some sort of automation, it’s something that is as updated as you are regarding your business and is thinking as creatively as possible without the sort of the cost structure, the overheads and just the sort of the pain of managing a thousand human beings. Just taking that part aside but bringing in all of that intelligence and just allowing you to point it in the right direction.
I think that would be really, really cool. In an ideal scenario, I would not want to be CEO for the rest of my life. I don’t want to be the CEO of anything, let’s put it this way, until unless I have to be.
Well hopefully maybe in decades time or something, this technology, whether it’s through us or whether it is somebody who gets inspired or builds on top of the work that we are doing, that effectively people or humans or human CEOs are only brought in to deal with exceptions.
The way to make decision making is literally outsourced. It is the big ideas. It is the big problems that we handle. And I think that one of the very interesting graphs that I share with my class every year is this graph of the adoption of new technologies. And we basically start off with sort of fire back in the caveman days which took tens of thousands of years and agriculture came in which took another tens of thousands of years and then you had the bronze age that became mainstream in 2500 years.
Then you have the Iron Age that became mainstream in 1300 years. And you start zooming forward, you see electricity became mainstream in 150 years, internet became mainstream in 20 years, AI is moving at a rate, which is a hundred times faster than internet. If you’re looking at super intelligence, you’re talking about like half a decade.
And then the question that always comes in is how do companies deal with this? Because then this is disruption at the rate of- it’s a crazy rate. Would companies start redefining themselves every few years? Would they be changing their mission every few years? And the fact of the matter is that the chances are that they might be, because how do you explain the scale and scope of how Amazon increased its variety and selection of things?
It would have been unheard of before the internet that an online website can have I don’t know, 20 million, 30 million units of things they can sell. The chances that organizations, especially small organizations, will be going through post sort of AI becoming mainstream, will have to change very rapidly every two to three years.
And to do that sort of strategizing and that sort of management, no human being can take that emotional rollercoaster ride. It’s just impossible. You do need AI to then do that job for you and you only handle the big chunky items. And I think that is where the future is effectively headed. Maybe if not in five years time, maybe in 10, maybe not in 10, maybe in 15, but I can’t imagine that it will take longer than 15 years.
I could be wrong, famous last words. I think that companies will have to change at such a speed that no human being can keep up. And so it is very important that you do that partnership now so that you can keep up. I think that’s what I would want to do in 10 years time. I don’t want to deal with the nuts and bolts.
I’m fine if somebody else wants to be CEO of Strategize at that time. If they’re a better fit, fantastic. But I’m still around and the company is still around. I’d rather leave the details with the AI. That’s where I think it will go.
A.J. Lawrence: The role of the entrepreneur, the role of human in the process of business kind of arose as ability to kind of gather resources in a manner that allowed for outcomes that just happen to perform better than other structures. We are pointing to a point where as you’ve said, AI kind of plays around with, well, why do we need X? Some of it’s just these are our golden idols.
We do this because this is what we do. And it’s like, well, guess what? AI will slaughter. Yeah, I’m going to massacre my metaphors here. Massacre our false idols but give us new ways of doing it, which will probably, it almost feels like we’re going to go to this regression and mean where everyone will be closer yet there will be more variation and more extent.
I think talking about where businesses are going, especially as a business owner is looking at, we’re told to think 10, 20 years out, it’s like great. You can still, but I think the adaption and the ability to reevaluate how broadly you are thinking your longterm versus short term, it’s going to be more and more important that your longterm is not tactical in any way.
You’re very directional. This is what’s happening. It’s like your need to be able to adapt on the dime is going to increase as an entrepreneur, as a business owner, as someone in business and your ability to use tools, be used, be incorporated, et cetera. Hey Borg, we’ll join the Borg. God, there’s so many great sci-fi references we can play with there.
This is what you’re playing with. Let me come back to you and have you come back on and let’s put a little more guardrails cause I’m doing this on the fly and that’s not fair to you or to the audience. But let’s kind of talk about maybe future about what you see really is how this is going to evolve. Because we brought this up late and it is really to where I think a lot of people’s fears and hopes are around AI.
It’s like, what is it gonna mean for me? I think it’s unfair to kind of just do that right now, . So lemme take a step back and be a podcast host and not just a geek. What’s the best way someone in the audience can find out more about what you’re doing, reach out to you. Where should they go?
Hamza Mudassir: I’m pretty active on LinkedIn. I don’t do much of Twitter. So, LinkedIn is a great place to look for us. Look for my name, Hamza Mudassir, on Google. I write a lot. You’ll end up running into some of my articles and if you want to just sort of read more, there’s also our website, Strategize.inc.
You can contact us through that. Generally quite responsive. I’m not a snob so feel free to reach out, discuss your ideas. Not everything has to be a commercial sort of transaction. If you just want to, I have some ideas, I want to reflect on it. I’m happy to chat with you on that as well. So yeah, absolutely do reach out. I’m sure A.J. is going to put in the website link, the descriptions. Happy to hear from you.
And look, at the end of the day, a lot of what we are dealing with is very, very new. Even the sort of the work we were doing and we are still doing is extremely new. One of the things I say to sort of all the people who are willing to listen to me speak is this that in the age of AI, as A.J. you alluded to, the biggest skill you can have going forward is your ability to philosophize.
And philosophy is really useful. It’s a really practical tool if you are dealing with situations that you do not know of. The unknown unknowns.
Those are things that no longer are problems of just vision and strategy, but a very much challenge for philosophy. And that is something that even when we are building our own products, even the engineers have to philosophize what this thing can potentially do. And where do we put the guardrails?
And so yeah. Happy to go really technical in our conversation. We’re happy to go really philosophical, and again, are happy to speak to businesses of all shapes and sizes. We can’t take on everyone, but if we have an interesting problem that we can potentially solve together, we’ll be very excited to find a way to work together.
A.J. Lawrence: And we’ll have in the show notes links to his articles, his LinkedIn, the site, his business, the email when this episode goes out, and also we’ll throw this up on the socials.
Thank you. This was great. Thank you so much for coming on the show today. I really appreciate this and I cannot wait to go talk more. This is just so fascinating. Thank you so much for coming on.
Hamza Mudassir: Well, it’s been an absolute pleasure and privilege to join you, A.J., and thank you all to the folks who are listening. I hope we can do another podcast sometime down the road and maybe we can share more of the sort of learnings as we bring this thing to life.
A.J. Lawrence: We will, I will make that happen.
All right, everyone. Thank you. I know there was a lot. I know we were all over, but I think this is really important and so much is happening. So how you look at it and how you play, just take some of the practical ways we were talking about and play around with what you’re doing. Evaluate what you’re doing. Please reach out, check out Strategize.inc, look at some of the thinking. You should definitely read the Harvard Business Review article Hamza wrote or co-authored. Definitely worth putting into your thought process, have some fun playing with this because it’s frustrating, it’s unique, it is dangerous, but it’s fun right now. What’s going on.
That’s kind of what we hope for when we are on our journey. So everyone, thank you so much for listening.
Hamza Mudassir: Wildest of the internet days, but only for the AI. So enjoy. Enjoy when it lasts.
A.J. Lawrence: Again, it’s another new wild west. All right. I’ll talk soon.
Hamza Mudassir: Take care. Bye bye.