Mohammad Oli Ahad is the Founder and CEO of Intelligent Machines Limited, a Dhaka-based enterprise AI products and data science services company. Prior to starting Intelligent Machines, Mr. Oli worked as a Business Analyst at British American Tobacco's regional IT team, where he led a number of global and regional projects for 25 countries in the Asia Pacific region, including Australia, Singapore, Hong Kong, and South Korea—some of the most technologically advanced environments.
Intelligent Machines was founded in April 2018. Their first product, Retail AI, went live nationally in October 2018. The product is currently being used by bKash, the market leader in mobile financial services (MFS), for merchandising activities in their over 300,000 active outlets in the country. The company has since expanded into multiple markets, launched multiple products, and is currently working with a growing number of organizations in Bangladesh and abroad to build AI and data analytics products in the retail, telecommunications, fast-moving consumer goods (FMCG), banking, and non-banking financial services industries.
I recently had the opportunity to speak with Mr. Oli about all things Intelligent Machines, artificial intelligence, and business building. We talked about a wide range of topics, including his personal journey to what he is doing today, how Intelligent Machines came into being, what motivated him to start the company and how he put together the initial resources to get started, what the early days of Intelligent Machines were like, the state of Intelligent Machines' business today, its products and services, strategic direction, and future plans, the state of developments in the artificial intelligence space, the evolution of Intelligent Machines as an organization, the challenges and rewards of being a founder, his evolution as a founder, his favorite books, the lessons he has learned along the way, and much more. What follows is a lightly edited transcript of our conversation.
This is a fascinating conversation in its entirety. I hope you enjoy reading it as much as I enjoyed doing it.
Mohammad Ruhul Kader
Thank you for agreeing to this interview. To begin with, can you please introduce yourself and tell us about your journey to what you are doing? You may also tie in your journey with the origin of Intelligent Machines — how did you come up with the idea and what was the motivation? Share a few reflections from those early days.
Mohammad Oli Ahad
Thank you so much for having me. We are Intelligent Machines. We build enterprise AI products. We have been doing it for about four and a half years now. We have deployed 32 AI models for 10 clients in four different countries. Any organization can come to us to get an AI solution to help reach its organizational goals, achieve strategic priorities, and/or mitigate challenges.
About me, my name is Oli Ahad. I graduated from IBA in 2006. Immediately after that, I started working at British American Tobacco. I was there for more than 10 years. For the first few years, I worked in the marketing function. That role gave me a privileged opportunity to walk in the streets and observe people and the world from 8:30 am in the morning to late afternoon at 4 pm for several years. The role allowed me to meet and observe all types of people, how they operate, what moves them and what does not, and how they think about the products that they consume.
I had rich exposure to the intricate workings of trade marketing. I gained a great deal of insight into the entire value chain from factory to warehouse to the distribution house to the entire massive distribution mechanism.
I had a unique opportunity to see how people actually perceive different discount offers, marketing campaigns, celebrity endorsements, and so on. Whenever there was a mega campaign from a large consumer goods player, I had the opportunity to see the consumer reaction in real-time from the market. When an organization designs a campaign, I believe they do it with certain purposes and based on certain homeworks. However, what I would often see was a disconnect between the expectations of these organizations and the real response from the customers. It was powerful to see. Most of the time there was a great disconnect between what the organization hoped to achieve — emotion and belief that they wanted to invoke in the recipients — and what the recipient ended up believing.
I still remember, in 2007, when I was in my trade marketing role. I was sitting at my desk in Chittagong and looking at some of the research projects that BAT was working on at the time. I clearly remember that the project was called Garden of Innovation. And as I was looking at the Garden of Innovation's questionnaire and the way the research team was approaching various consumer problems and analyzing the data, I could clearly see in my mind where this organization would go in a few years and how the competition wouldn't even be able to imagine how to catch up.
That's exactly what happened. Today, the phenomenal growth and success of BAT that you see, at least in our country, I would say, is largely based on those early research efforts.
After three years in that role, I joined the regional technology team at BAT, where I worked with a team of 32 people serving 25 countries in the Asia Pacific. My market included Australia, New Zealand, Papua New Guinea, Hong Kong, Singapore, Malaysia, Korea, and several others.
That role required frequent travel. While in that role, I was fortunate to lead eight regional and global projects for these different countries. During this period, I gained exposure to a number of things that I would say eventually led to Intelligent Machines. I could see first-hand the power of data, the power of IoT, and the power of advanced technology projects.
Anyway, during my work in BAT's regional technology team, I had firsthand and close exposure to the power of technology.
There was an important aspect of this experience that I want to highlight. I was working with teams from different countries. One thing that caught my attention was that teams from many South Asian markets performed differently.
Although I had the same budget for all the markets, surprisingly, none of the markets in South Asia such as Bangladesh, Pakistan, Sri Lanka, Maldives, or Afghanistan, ever requested original projects. The most common practice here was that the head of marketing/the country manager/the head of operations, or the finance director, would usually ask if there was an existing solution for their problem from markets like Singapore, Australia, or South Korea and request that.
On the other hand, my colleagues in those markets would never do that. They would almost always say, "Oli, we have these challenges, and we need resources to design a solution."
The challenges for the head of marketing in Bangladesh and the head of marketing in Singapore would be the same, such as how to reduce mistakes in memo writing or how to ensure the most robust demand forecasting mechanism or the finance team would say, we have a rolling multi-currency cash flow forecasting system, but we would like to have a more accurate one.
That’s such an interesting and I think somewhat accurate observation.
The amazing thing was, I don't know whether it's cultural or because this is how these teams have been doing things historically, but my colleagues in Australia, Singapore, Hong Kong, and South Korea would generally ask for a project and that would mean that we would get down to the problem, break it down, and go deep into it. Then from there, we would sketch a solution from scratch.
On the other hand, in markets like Bangladesh, they would ask, "Do they have a solution for this in Singapore? Do they have a solution for this in Australia and South Korea? Show us that, compare it with our situation, and then tell us the cost." This was bizarre and weird to me.
I remember writing a detailed document around 2010, the charter for marketing ideas. It contained 32 solutions that should be built over the next two years for Bangladesh. I launched a massive personal campaign and approached the Head of Marketing and the Managing Director. I said we have the same budget as every other market, according to the global rule and the regional guideline. So with the same budget, we should build from scratch. After completing the business analysis, we call for tender, and then IBM, TCS, Wipro, HP, and sometimes Huawei and HP pitch. But we could follow the same approach as them. However, I was never able to convince my colleagues here.
Once again, I started to feel a general frustration. I thought this was the wrong approach. It seemed like a forced fit.
In technology, there are three categories of solutions from a fitness perspective: one, fit for purpose, which means the solution matches what you need and what you are looking for; two, fit with changes, which means you have to change your processes or your behavior to make the solution work. This is what Bangladeshis usually did. And three, change requests, which means you modify the solution or the product itself.
My colleagues in other countries would either do fit for purpose or a change request. They will always fight if you request them to do a fit with change — this is the solution, can we update our policies and our practices so that we can fit with the product? Generally, they would wage a war.
Unfortunately, here, it started with a fit with change. A head of functions or a director would take a product and then command his team: "Here is this product and for this, we now need to do these things in this way." I believed it didn't serve the organization's interests or the team's interests right. That was one personal agenda that I always dreamed of: if I could have a team who would look at our problems and build a solution from scratch.
The second one was a bit more personal. Many of our colleagues, classmates, and relatives started migrating abroad. I noticed at that time that people were leaving the country and going abroad. As it happened, both my wife and I were firm on staying in Bangladesh, but we started worrying about whether we were making a mistake, not doing what was right for our children. Because all of our classmates and relatives who moved abroad, one of the primary reasons they gave was that they were doing it for the future generations. Naturally, it got us thinking about whether we were bad parents, whether we were not thinking about the future of our children the way we should. One good thing was that because of my role, I traveled frequently.
You are already thinking about your next move and I’m assuming these discussions would eventually lead to Intelligent Machines.
What I did was make it an agenda. The next time I flew to Australia or Singapore, I would gently request one of my Bangladeshi colleagues there to share their views with me. I would tell them that I had this question in my mind and ask them for their advice. Sometimes, we would have this conversation over breakfast or dinner. A strange thing happened. Remarkably, I started getting only one-sided advice. I honestly expected there would be differences of opinion, or maybe I didn't know what to expect. But when this conversation started, I realized that I had a cognitive dissonance. I possibly expected some of them to say they were very happy and some of them to say they were not very happy. But I never imagined any of them would say they were unhappy. I didn't expect that.
Unfortunately, I don't know whether it was with my colleagues there, but again, because of the different respondents from different countries, different economic environments, and different cultural settings, that somehow evens out the bias but again, perhaps the organizational bias was there, I don't know.
Anyway, they started telling me more from an economic perspective, yes, they feel thank God, they are a lot better. But would they make the same decision? All of them said they would not make the same decision and that was a very powerful thing for me. Personally for me, and my wife, I would say, that was convincing for us and we concluded that we don't need to have this discussion anymore.
Many of my colleagues, who had lived there for a decade or more, were very senior graduates from BEUT or Dhaka University. They were outstanding students who ranked high on the merit lists. However, they shared their personal opinion that they could not provide the best environment or opportunity for their children. Unfortunately, they felt like they couldn't do that. They admitted that they gave them better economic opportunities, but not a better life in terms of environment and other factors.
In these discussions, a natural question that arose was why they had decided to migrate in the first place.
They gave a number of reasons, and I believe all of them were valid. I realized that we could not address or mitigate most of those reasons in a short period of time or within one organization's scope. Let’s say, Bangladesh as a country could not provide adequately what these talents were looking for.
However, there was one area that caught my attention and presented a huge opportunity. They all agreed that there was no place to utilize their talents, to offer them the learning and growth opportunities that matched their intellect, capability, and potential.
I felt that, like customer interviews, these people told me one thing, but I also saw different things happen. They claimed that this was the situation, but as an interviewer, I sensed that it was actually the opposite, based on my experience. I also noticed that the heads of functions in my organization had the same budget of GBP 10,000 for every quarter, but they were not utilizing it. Yet, they told me that there were limited opportunities for talent. I thought these two issues could be connected.
These were the initial points for our business case at Intelligent Machines. As I thought more about it, I wondered if I had confirmation bias, because I started seeing other connections. One of them was the inspiring story of our readymade garments industry. A unique selling point of Bangladesh was that we could provide superior quality at an unbeatable price. I felt that this could also apply to tech, as we could offer superior quality at an unbeatable price since our living expenses are lower than in many cities around the world.
I learned from my colleagues in BAT that renting a small home in the Netherlands or Australia was very expensive, more than BDT 300,000. Moreover, the tax rate in the Netherlands was very high, around 49% to 51%. So even if I earned 600,000 taka, I would only keep 300,000 or less after paying taxes and other expenses. I had to be very careful with my money even with a 600,000 taka salary. My colleagues and I used to share our personal challenges with each other. I was familiar with the lives of my colleagues in the developed markets. Many of these projects, although they were for the Asia Pacific market, had developers from Romania, Poland, Germany, Netherlands, UK. I learned about their situations. They were senior or mid-level managers in BAT, earning a good salary. I thought they would live comfortably, but they told me that they had to operate within a strict budget.
So I had another point to add to my business case. We could offer superior quality at an unbeatable price. There was no way a team in Europe could match the salary that we offered here and still have a happy team.
As people age and mature in life, their perspective starts to change. I experienced this myself as a father. My elder daughter, Armela, was born on October 22, 2008. As she grew older, my wife and I used to share stories from our work with her during our dinner. We thought that by doing this, she would naturally start to build ideas about leadership, teamwork, perspective, and long-term thinking. But when she was around four or five years old, I noticed that I was giving fewer and fewer examples from my workplace. I didn't want her to know that I was working for a product that was harmful.
I started to question whether I should change my industry because I felt that I was not being a great example for my daughter. However, just for the record, I love BAT. I'm so deeply grateful for my years in BAT. I still love all my colleagues and line managers, and I owe a lot to BAT. I am who I am for a number of reasons: for my family, for my friends, for my teachers, and also because of BAT. I learned how to take care of people and be sincere in whatever I do. I would say BAT groomed me in fundamental ways.
However, as I reflected on the product and the impact that we were making on the community, I began to feel that perhaps it was time for me to explore a different industry. It seemed inevitable, and one day I simply handed in my resignation.
What happened after that?
Shezad bhai, who was our Managing Director at the time, asked me to come and meet him. He asked me what I was planning to do. I told him that I wanted to open up a technology team, build a technology team that would provide advanced big data analytics and AI solutions, etc. Back then, AI was not a big thing yet, so it was mostly big data analytics services. Shezad bhai said, "Okay, go to him, go to him, and talk about it." Later, when I started, I realized mashallah, and how privileged and blessed I had been because of my mentors and alumni. They actually talked about me in the industry.
One thing that they loved about me a lot was my combination of business background and deep passion for technology. And I always championed business, and never technology for the sake of technology alone. I never cared how fancy or fascinating, complex, or advanced the technology was. I was always keen on solving the business problem with the simplest and most effective solutions. So, from their feedback, I came to know that they valued that a lot. So when we started, mashallah, within the first two weeks, we got 13 work requests from companies like Daily Star, Aarong, BRAC Bank, NTV, Apex Footwear, and ACI. I just took a desk at Moar. We never told or called anyone.
Every project was fascinating. Apex wanted to have a better forecasting system for their shoes, which design styles would do well. Nasim Manjur Elahee, CEO of Apex, said that every year they struggled with one problem: before Eid, they tried to understand customer demand as much as they could, but unfortunately, every year, the scenario was the same. Some of the new pairs of shoes became hits so quickly that they ran out of stock, and the rest just piled up. They wanted to know if there was a better solution to forecast customer demand.
Aarong wanted a solution to better understand their supply chain.
Each problem was powerful. For instance, Daily Star wanted a customized and personalized reading experience where the AI would know the reader's preferences, such as simple or sophisticated words, short or long sentences, related stories or quick summaries, graphs or text, and even the size of the graphs. Then the AI would offer points based on the reading style and suggest events and other opportunities.
Each of those 13 projects was fascinating. Anyway, Allah knows best. For a number of reasons, the one that we picked was a project of bKash. bKash's finance director Moinuddin Mohammed Rahgir bhai, who coincidentally was also one of my assessors in BAT. In 2006, when I went through the interviews and assessment process at BAT, Moin bhai was one of the judges at the final presentation. It was very tough and somehow during the presentation, we felt like we clicked.
As it happened, I always looked up to him as someone who was very modern. He was quite senior to us. We admired him as someone who was senior but also very modern and smart. He was much smarter than us. But I rarely talked with him. I had barely any interactions with him. Maybe once or twice in my entire tenure at BAT. Again, I’m so grateful to all my peers who spread the news by word of mouth. So Moin bhai called me. I still remember that day. I was in a car crossing what is now SKS Tower. Moin bhai called and said, Oli, I'm worried about one thing. At bKash, we place so many posters, stickers, festoons, and banners. I wonder whether we're making the best use of our resources and whether we should use fewer posters and other POS materials and more customer discounts, cash backs, and other types of consumer engagement. Is there any way that data could tell us what would be better? Can your team do that? I said bhaiya, can you arrange for me to get regular images of these outlets? We would have a mechanism where you would pay a team who would send us photos of these outlets, and then we could give you a correlation and insights. He told me, Oli, just send me a mock-up. I said I'll do it in PowerPoint. He said, whatever, just send me a mock-up. So I went to Moar Banani and took a photo of an outlet from the street. Then I marked on the photo how the data would work and sent it to Moin bhai. Moin bhai forwarded it to their commercial trade marketing team and they asked us to make a presentation.
I said no. Before anything, I wanted them to give us permission. I would invite some of my juniors from IBA and we would visit their outlets and markets. They could not ask us anything or restrict us in any way. We would have an open license. They would not be able to stop us from asking or doing anything. I wanted this before anything else. I would not go and make any presentations.
They somehow liked the idea. They agreed. So then we had four types of outlets/markets: rural, urban, semi-urban, and industrial markets such as Gazipur where there were lots of garments and others. We rented two microbuses. I invited some of my juniors from IBA. Again, I’m very grateful to all of them. I don't think we were able to pay them anything other than lunches. That visit was so rich and powerful. We went there. We made videos. We talked with the teams, the retailers, and the consumers and then we saw the challenges that the bKash team was actually going through. A number of those I would say such as they had more than 80 different types of POS materials, but no sensible way of managing those POS materials in execution.
We visited several distribution houses and found stocks of POS materials in closets and rooms. Unfortunately, the colors were fading because they had been sitting there for a long time, while the neighboring territories were crying for POS materials. The problem became so severe that the retailers often held the sales reps hostage, saying they would not let them go unless they gave them a festoon. Then the retailers started making all sorts of accusations in their minds, thinking that they were being discriminated against because they were small retailers and that the sales reps must be giving the POS materials to the large outlets only. But the sales reps honestly did not have any. And the retailers did not believe them, how come you're doing such good business, everyone asks for bKash and you're saying you did not even have a festoon after three months? And the next distribution house had plenty of them and claimed they had no space to put them. That was just one problem. The other one was the poor merchandisers who had to carry and put these POS materials in outlets, and who had such a long list of work to do.
We made a video, I remember, and we brought it down. We did a very basic time-motion study and showed it to the bKash team in our presentation. We said, see "Your team member needs to do 60 different unique tasks just to execute one POS material. I mean these are tasks like putting down your bag, getting all the materials, sorting them, thinking which one would fit here, and this is how much time it takes. Right now you're telling them to do X number of works, but ideally they wouldn't be able to do even 0.4 of X. And because of this disconnect, they never hit their variables, they're always underpaid and they're always demotivated. It's a bad cycle, unfortunately." When they saw these insights, they were like, we want this team to work. We didn't have anything. We didn't have the technology, we didn't even have a team. We formed the company on June 6, 2018, and on July 31, 2018, within 25 days of our inception, bKash gave us a large work order of $900,000. I still remember when I took that paper. I was in Rashed bhai's room, who was the head of supply chain, procurement, and media buying. Multiple storms were going through my head. The first thing I honestly thought was, "Is it really clear? Did we make any miscommunication? Do they really know that we don't have any product, we don't have any office?" The second thing was that this is bKash, they are so gutsy and so bold, and they are giving this chance to a fresh team with zero portfolios. And then I thought, "Who should I rush to? Where should I go now?" Because the previous team, my dear brothers from IBA, were all business students. None of them that I knew were capable of coding. I was the only one who I knew could do coding. But I knew that I wouldn't be able to do it on my own. It would take months.
I was in a rush to find someone. I met Aman, you could say he was the first team member of IM, in a French language class at Alias Francis Dhaka. He was a finance graduate from Dhaka University's business faculty.
We were all sharing why we came to learn French. My reason was that, at that time, I had taken my daughter Armila to enroll in a French class, and I thought it would be wonderful if I could also speak French. Then, father and daughter, we would have fun speaking a new language. I also enrolled in another class and met Aman there. The reason Aman gave was that he had always wanted to be a footballer, but unfortunately, he had torn one of his ligaments and could never play again. The next best thing for him was to become a manager of a European football team, and to do that he wanted to learn multiple languages to improve his chances. He was also learning data analytics to be a nerd. I was blown away by Aman. He reads a lot and usually gets into deep conversations about philosophy. I don't remember whether I called him or met him, but I told him that bKash had given us this work and we were looking for teammates. Aman went to one of his friends from DU IT named Akash, who then met me. Akash brought in Manosh, and Manosh brought a few more people. That's how it started. This was the story. But I would say that, looking back after four and a half years or so, I feel extraordinarily lucky in many ways. One, we got such a good client as bKash, who worked at an amazing speed, which was super helpful for us. Two, we got such a talented and passionate team of people who made this project possible.
A startup team, especially if it's a b2b startup, would benefit greatly from having a fast-moving client, a client that moves with speed. That is the ideal scenario for a startup, in my opinion. For bKash, not only are they fast, but they also have phenomenal resolution. In all these years, we have seen that if bKash says they will launch something on January 9th, they will do it 99% of the time. They are serious about their deadlines and that helps us to keep up with them. Unfortunately, I have to say with respect that the general corporate scenario in our country is not as professional or capable. I mean it in a respectable manner.
I'll go on a tangent here. I request you to consider sharing this. Working with these companies as a business partner has given us this exposure. Since childhood, we are used to complaining about government workers that they are slow, lazy, and inefficient. That was the perception I had from magazines and other sources. It was very naive of me. I thought the corporate world would be completely different.
The reality, however, is different. I want to tell all our respected colleagues in the corporate world that Bangladesh needs to recognize this: we are far behind in areas where we cannot afford to be. I share this with care, empathy, and respect. You cannot take their deadlines literally. You always have to consider the context. If they say they will launch this in the first week of March, or they will review this and give a decision in April, you have to be skeptical. It is like a 3 pm meeting that will start at 3:30 or 4:00 pm. Unfortunately, on the world stage, you can never take things at face value. And I wonder if we can become a very developed and capable country unless we improve these things. We don't take our own words seriously. When a client or a team tells you to give them a POC, and they will review it and do something, sometimes I doubt that they believe their own words. Because this is the pattern. It is not just one or two cases.
A refreshing exception is bKash. In most cases, when they say something, we as a team know that they mean it. Again, there are also differences among different teams there. But for most of the things, especially the teams that we work with on commercials and procurement, they generally stick with their words and they mean it. I know I went on a tangent, but I would request as someone who is a b2b provider to all our industry colleagues and peers that in terms of being more responsible, being more agile, I don't know how to put it, I would request our private industry team members to really ask themselves these questions: are they really operating at the most professional capability level that they can or they should? I believe there's a long way to go. So I will say definitely, we were very privileged, we had bKash as our first client.
Let me quickly give you the next few bits. One day, the Unilever team invited us. Shadman and Parvez bhai, who both headed two different verticals of consumer research and market research at the time, invited us. We were so lucky to have Shadman and Parvez bhai. They showed us a problem that “in our consumer engagements, one of our key objectives was to convert target consumers from competing sources of business. Until now, we had been doing it this way, but they asked us if AI could help.” The job would require reading handwritten shop receipts and other things. So then, mashallah, another very exciting and fascinating journey began.
So, when we started all these projects, we did not have any formula or any method. Usually, what we used to do was to search and download the latest PhD papers and thesis papers whenever we had a new request. We wanted to see if any of the researchers anywhere in the world had worked with the same problem, such as reading handwritten uneven scripts. However, we never found a solution. We only learned about some of the problems that those researchers were facing. For instance, in this case, one problem was how to teach the computer when a sentence begins and when a sentence ends. When you write a letter, how does the computer know where the sentence starts and where it ends? And even in a tabular format, if you train the computer to read the texts, how does it tabulate that data? How does it know where the rows and columns are? These are major engineering challenges.
One major challenge we faced when we started our project was the diversity of writing styles among people. Moreover, Bangla has many graphemes, which are the smallest units of handwriting in a script. We began by collecting shop receipts but soon abandoned that approach. Then we went to Katabon, where roadside stalls sell old newspapers and various other papers. We bought many kilograms of those papers but still made no progress.
Next, one of our team members suggested that we visit Ideal School in Motijheel and talk to the principal teacher. They were very kind and gave us 20,000 final scripts of class 9-10 students without asking for any money or anything else. It was a very generous act on their part. Our team members entered those scripts into our system but again, we did not get any results. We realized later what we did not know then: that there was a selection bias in our data. Most of the students wrote in a similar and decent manner, which did not match the style of the shop receipts. So when we were taking that learning to the shop receipts, it wasn't working. Our system worked for the scripts but not for the receipts.
One day, a team from BAT visited our office. The team was led by Golam Sakib Choudhury, a junior alum of both IBA and BAT. Sakib and his team asked us what we could do to solve their problems. One of the problems they had was how to verify if the sales reps were delivering the right brand message. It seemed like a problem that AI could solve. However, when we started reviewing the research papers, we realized that all the successful speech recognition solutions at that time could only handle up to one and a half minutes of speech. This was sufficient for voice assistants like Siri, Google or Alexa, which only needed short commands. For example, "Hey, Siri, what's the weather?" or "Alexa, what is the recipe?" But our challenge was different because we had to analyze five-minute clips and provide the results with the same accuracy. This was again a different ballgame.
We aimed to deliver our results in 0.7 to 7 seconds, a number that we derived from various factors. This was the time limit for the AI to complete the analysis, and the existing mechanisms were inadequate for this task. Alhamdulillah, we had some highly capable team members. I would like to mention a few names: Omar, Shaklin, Gazi, and Mashroor. Shaklin had a different interest; he wanted to explore game-making, so he took a leave. Gazi worked on script recognition, while Omar and Mashroor worked on the handwritten text. We were very fortunate. Customers came to us one after another with their problems. The IDLC team approached us and asked us to build an AI for them that could process CIB reports and bank statements, a problem they were facing challenges with.
We have been fortunate and privileged to start working in an area without realizing that it is a supply-constrained market. It was like a few years back when CNG was in high demand and customers would beg them until Uber came along. I hope we did not behave like them. The reality is that we are operating in a supply-constrained market. Even today, if you search on Google, you will not find many teams who would make you a tailor-made AI product.
There are possibly hundreds of or more teams who are working with many different AI products. But you will probably have a very hard time finding a team to whom you can go and make a request like a tailor: I have this problem, can you make a solution for it? Because it probably does not make sense. Probably all the other teams are thinking AI is a lengthy endeavor. So they would make one product and then sell it to different customers. Probably this particular business model does not look very appealing to most: that we would have a shop where anyone can ask for one AI model. Probably that is why this area has not taken off generally. Since we have been operating in this area, our customers have recommended us to other people.
One day, a Bangladeshi brother who lives in Australia contacted us. He was a senior graduate from BUET and had been living there for a long time. He asked us, Oli, would you be willing to make an AI service for a mining organization in Australia? I wondered and asked him, How did you hear about us? He said he had asked around and people had referred us to him. And others would tell us that he had met a teacher at NUS in Singapore who had heard from one of his Bangladeshi friends about our team. There were usually two or three points of connection. That is the background story.
This is wonderful. I would also like to talk about another background story. You started in June 2018 and you had a wonderful roster of clients who came to you. That was the business development part of it. I want to learn more about the making of IM as an organization. What were the challenges in the early days? How did you put together the team? What kind of organization was it in the early days? How much has the company evolved over these years? What are some of the lessons that you have learned in this journey? Please share with us the history of the organization.
Thank you so much. It's a very painful story as well. We made so many mistakes. I feel like I have exhausted my quota of mistakes and I have made four or five times more than what one person would be allowed to make. Again, I would say, all praise is for Allah only. And I'm immensely thankful to the team for supporting me throughout this journey. It sort of evolved organically. I was only the curator. I just ensured that there were a few firm policies, that's it, nothing else — these are some things that we wouldn't do, etc.
As I told you, what happened was that I reached out to Aman, and he brought some of his friends. They, in turn, brought some of their friends, so this is how it happened. It happened in two ways. One way was that an existing team member brought in another team member. For example, Imran brought in Apu, and Asif brought in Shetu. This is how it generally happened. Team members who were working with us brought in their university colleagues or classmates. For instance, when Omar joined us, he was still a student of IUT. His classmates learned about us from him and wanted to join us as well after learning about our work and culture. We always got this feedback that people found our way of operating inspiring and appealing. Omar said, "Bhaiya, some of my classmates and juniors are interested in working with us. Can we invite them?" And this is how it usually worked.
The other way was that people proactively reached out to us by cold emailing or messaging us on Facebook or LinkedIn. We actually had a large part of our team members through this. For instance, Imran wrote to me on Facebook. He was working at Grameenphone back then, after completing his graduation from BUET. He wrote to me on Messenger. I said, "Sure, let's talk, but I'm curious. You're working at Grameenphone. We wouldn't be able to pay you that much or give you that fancy office or any other benefits. None of the benefits would be there." He said, "You know what? I joined here believing that I would be able to work with data and AI. But unfortunately, opportunities seem to be very limited." And this is how he came and joined us. Similarly, Adib came to our office one day after a brief chat online. He was a student of naval architecture and marine engineering at BUET and was passionate about tech and AI.
This has been a common pattern for us. Many times, many of the team members had zero relevant engineering experience. Adib came in and said he was interested in our work, but he was studying Naval and Marine Engineering. From the beginning, this was how I responded to people. Perhaps I took this lesson from BAT: I don't care what you can do now, I care how you will behave in certain scenarios, when things get tough, when things get very stressful. Will you become a tyrant or break down? I never cared about credentials as much as I cared about how you would behave. This is what I learned from BAT.
In BAT, we were taught that you first ensure the cultural fit, then you check for the technical fit. Without ensuring the cultural fit at BAT, you would not start to see how well someone knows marketing or finance or others. You start with preliminary interviews and then go to the assessment center, you check their behavior and there are large points only on behavior. Does this person try to convince others when he knows he's right and the others are not seeing it? Does he care enough? What are the tools that he uses? And does he update his or her own perspective, his or her own views? In BAT, I was a certified assessor and interviewer. We all went through that same interview training. And that's where I learned this: you don't ask how great someone is at his work if your organization does not function well.
There are two different types of organizations. One type does not care about the team camaraderie. They just want brilliant, powerful generalists. They do not mind if people hate each other or fight with each other. They are individualists. That is one type. BAT was not that kind of organization. So the first three steps before the CV screening, at least the two interactive steps: the preliminary interview and the whole day assessment center, did not involve any questions about whether you knew engineering, marketing, finance, or whatever role you were in. Everybody had to go through the same assessment and these were all behavioral. So at IM, either one team member invited other team members or people approached them through LinkedIn and Facebook. But we always maintained that we hire for who you are, not for what you know or how smart you are.
Having said that, it did not always work out and I would say that was part of the process. For example, one sister came in, she completed her graduation in architecture. Then she went to a government institute to learn Android application development and she was with us for almost a year. However, we did not see noticeable improvements in her performance. So, again, many things happened. We gave many team members opportunities, and many team members gave us opportunities. I would say in most of these cases, we eventually came to realize that we were not a good fit.
Here are a few of the principles that helped us. One, we are looking for team members who would ask questions and who would expect questions. So people who were not used to sharing their views and problems might have a problem. It could be a challenge in our environment for someone who was not comfortable with this and might find this stressful. I recognize that. Another principle is zero supervision.
We have received many complaints from team members that our current system does not suit everyone. They often request a mechanism where their supervisor would guide them and monitor their progress. Personally, I am not interested in this, I do not enjoy supervising people. I don't fancy supervising people. This doesn't just interest us. We usually look for team members who are self-motivated and want to do a lot of work and would feel guilty if they do not deliver. Unfortunately, we have seen some team members, who had previously won awards, participated in multiple programming contests, and so on, but then with us, they did not write a single line of code for three months in a row, because the environment is such that it doesn’t support their way of working. It's counterintuitive
As a startup, we have very limited resources to do many things. Therefore, you might expect that we are keen on ensuring everybody's full utilization. Unfortunately, what happens is that even the leads are also writing code and developing solutions. They have regular routine meetings, but these are short meetings. Our team meetings last 15 minutes. We encourage people to have as few meetings as possible. So generally, you don't have meetings. Only once a month or so, we have all hands. This is when we take longer, one hour to one and a half hours. So if anyone needs anything, unfortunately, they can spend some more time without sharing their problems. The leads already have a number of problems and challenges on their own, and we don't always do a good job of supporting everyone. I know in the back of my mind this feeling: is Mayisha doing enough? Is Nazmul doing enough? I know I have this problem in my mind. But I don't get to act on it. So what we do is we start by looking for people with the right behavior. We want to be with people with whom we would like to spend time and be together. That is one key point.
Second, we always want hungry people, people who would get up and do the same things repeatedly until they are satisfied. We routinely discard tons of design works after many revisions because we felt they were not good enough. That is what we do.
When it comes to learning, there are a number of things that we do. In all hands, we always share everything transparently with the team, such as the bank account status, the transactions, and so on. There are a few things that you need to know: how long the organization's cash flow will last, till when our salaries are secure because we believe we have a responsibility to our family as well. If I need to look for jobs in November, I need to know it now. If I don't see any prospect of a salary review, and the amount I'm getting now doesn't match with my family's requirements, I need to know it. We discuss these things openly. We used to share our salaries with everyone. I even wanted to have them publicly published, but then I realized that not everyone is comfortable with that in our cultural norms. So, we took a step back. But still, almost everyone knows everyone's salary here.
There are at least 15 team members who always know every single transaction in the company. We have signatory A group, signatory B group, and signatory C group. For most transactions, signatory A and B will handle them. I only get involved in one or two transactions in a year, usually when they exceed 5 million taka. Otherwise, I don't. I always see the results, see the reports, and I'm updated, but I don't get involved. We have regular cash flow discussions where everyone participates and has a fully transparent picture. In leads meetings, we have a leads team where we try to discuss all the decisions.
We have a very free-flowing, flexible, and transparent arrangement where the understanding is that if there's anything important, someone will inform everyone. For instance, if Moar is increasing the rate, or if we need to do anything, or if the NBR is asking for something, the person who's dealing with it will write in Slack or WhatsApp to all the team members, and everyone will get to know and at some point, we will come together and discuss. That's it.
As for what we learned, I would say that we are much more careful now because it hurts a lot when you discover or realize that you have people who are not a great fit and then you inevitably have to have that conversation. It is very painful. We try to avoid this now. Based on that lesson, we start with a more informal arrangement now. In the beginning, on the first day, we started with a full-blown contract. We talked with you, done. We started full-time right away. The concept of probation or any sort of testing was alien to us. In fact, we were strongly opposed to it. We thought that if we agreed, we were best friends from day one, just like in childhood. Seriously, that was the case. We were best friends from day one. And then everyone would know everything. But then we realized that the reality of life is that people can be and are different.
Now what we do is, when we start working with someone, we request that person to talk with as many team members as they can. Because if they talk with only the leads, maybe we would only give them things that we know. Maybe we wouldn't even know all the things that are going on in the company as leads. Maybe people are not giving us the hardest feedback that we should know. Maybe we will not be able to give them the true picture. So we give them the phone numbers of all the teammates, and we invite them into our Slack. There, they get everyone's phone numbers, and then we request them to make sure they go out for coffee and talk. If they talk with more team members, they will tell them how the culture is, how people treat each other when times are hard, etc. We encourage the potential team member to know about us as much as possible. I would say of late we have had far fewer such experiences.
Another good thing is our internship programs. Our internship program has been very, very successful for us. I mean, most of our full-time team members were actually our interns. We take interns in large numbers, and from day one, we try to offer them good pay. We generally offer them BDT 30,000 but the thing is, even as an intern, you can earn up to BDT 170,000 and many team members actually did earn more. So we never differentiate between interns or part-timers, we always see the value that they bring in. We had an early discussion, years back, if people come to see a match because of Messi, should Messi be paid the highest, or the manager be paid the highest? We discussed and agreed that you know, since Messi's bringing the highest value, he should get it. Many of our team members' salaries are actually higher than their leads'. I mean, it used to be the case. These days, I would say that's not anymore. One other thing is that 100% of our engineering team members go abroad in three years or so.
Currently, we have a few team members who fall outside of this. Previously, this was almost the norm. Therefore, we always hire interns, which gives us a good opportunity to assess their tenacity and commitment and they also get to know us. From the interns, we then offer full-time positions. This could be a reason why we have almost no experience of the kind that we used to have of late.
These are interesting experiments.
To summarize, one important thing was our recognition and realization of our limitations. Earlier, we didn't realize we were not great line managers. So what happened was, we didn't filter out any team members, which caused all kinds of frictions, such as you're not seeing me, you're not helping me with where I'm getting stuck, and all those. Now we know that as an organization, in our culture, as of now, we have this limitation, we're not great at supervision. So we tell this 10 times/20 times before hiring someone, please think about your student's life, were you someone who used to regularly need teachers' help, or parents' help to navigate challenges? If so, then we might not be the right fit for you. Some of our old team members hated us and made scathing comments about us because we couldn't help them while working for us because we are culturally different.
Now we realize that we won't be able to give you this important help. This is important. But unfortunately, as of now, we don’t have the means to give you this help. If you are someone who comes up with solutions on your own, we would be able to help you. So the realization of this, our limitations and then proactively communicating this with others and then trying to filter out intern team members using their past performance such as which of the team members always needed to follow up, don't invite them, which intern team members didn't need as much supervision, invite them because we serve their requirements better. That's it.
This realization has been extremely helpful. Earlier, we didn't have it. I will tell you that for most of our colleagues we are their first professional experience. So our responsibility is even greater. I feel so bad for some of our early team members who didn’t have a good experience with us. We need to tell them what professional world looks like. It feels like another important responsibility. I believe my other lead team members would agree that this has been a key differentiator when we realized and acknowledged that this is important and this is where we fall short as of now. We don't see any possibility of having enough financial capability to have a proper structure to offer guidance to people.
As of now it is just very flat: only one lead for 15 AI engineers or 17 data scientists. It's not possible. So we realized that and we made it a point to tell this again and again. We ask new hires/team members to talk with the others and ask questions to learn more about our culture and limitations, chances are they would also share that don't expect a lot of help and handholding. We would have loved to provide better mentorship to new people, but we couldn't do it at this stage. Mentorship is one area where IM lacks significantly. We tried different things. We tried onboarding senior advisors from the USA, and NRBs and we paid them, but it didn't work. We realized that we have our limitations because of the number of tasks, the project cycle, and the project's nature. This is one area where we fall short. And this recognition of our limitations has helped us have fewer bad experiences these days.
So now you are more conscious about the kind of people you hire — you hire people who manage themselves well, instead of hiring people who need constant supervision or support to operate. That's interesting. A few more questions about the journey. One question is, how did you put together the initial resources to get started? How big was the team at the time? And finally, to connect these two, in what ways has IM changed over these years as an organization? You touched upon a few major changes, such as the growing realization of your own cultural reality as an organization and shaping your recruitment process accordingly.
To answer your first question, it was very organic and a bit strange as well. As I said, I reached out to Aman, and Aman reached out to others. That was exactly the case. Only once did I get very desperate. I was looking for a product manager, so I wrote a post on Facebook or LinkedIn, I copied it everywhere, explaining that we don't actually care much about your academic background or experience, we are looking for people with these behaviors and traits. Would you be willing to be a product manager? From there, I got Shubho, and Charles Gomez, and apart from that, there were never any job classifieds or anything.
When we start a hiring process, we only take your email address.
We are a resume-free organization to the extent that we wish not to know your school name or whether you're a man or a woman because we want to be as neutral as possible. So we only require your email address. Consequently, we never open any CV sent to us. It is a religious practice not to open them as we don't want to be biased. We only take your email address and then send a test link via Codility. This is the platform we use for the first test.
On Codility, we generally provide one easy/medium question and one hard question. The approach is to understand your thinking and problem design approach, and how you approach problem segregation. If you pass that stage, we send you other tests on workarounds, and then we have a meeting where we discuss and conduct culture assessments, among other things. In terms of the initial team, when we started with bKash, obviously, I was the only one. Then Aman connected me with Akash, and he brought in a few people, and we quickly became a team of six people.
How we have changed over the years is a hard question. I would say two things. One is change at the senior level, which is one important learning. Looking back now, you might wonder how this could be a learning; this should be obvious, even a fool should know this. But trust me, many times or perhaps most of the times, the most obvious things are not so clear, unless you experience them. This is about how I behaved in those early days and how the leader's behavior impacted everything.
I wish I had known this earlier. In those early days, I felt like I was just another team member. So one team member can act the way they want. I can also act the way I want. But that was not true. And that was very harmful to the organization. So whenever I used to get angry or upset, that had a huge impact. So one major change would be that I realized I couldn't behave like anyone in the company.
It is not about being authentic or not. I'm authentic. But this is one area where things have changed. I have learned that it is important how I behave. I have been more careful about how I operate. For example, now if I'm very upset about something, maybe I would first share it with only Shuvo Dep, only Rashed, and only Fahim. Earlier, I didn't care. Whoever was there, I shared with them. I talked with everyone. I was just like my youngest child Ornila would be; she's just eight years old.
I was just like that and that, unfortunately, was not good or helpful or healthy for my teammates and the organization. I used to get upset and then wear it everywhere. In all hands, if someone did something wrong or if someone couldn't understand something, I would be upset in front of everyone.
One lesson is this realization that since my behavior influences and impacts everyone in a disproportionately bigger way, I need to be careful. I cannot afford not to be careful. That's one major lesson I would say. Previously in our all hands, people would anticipate anxiously how Oli bhai would behave today, whether he would be upset or get depressed or sad, but now it does not happen anymore.
The second thing I would say is that we now have an organizational history and everyone knows how we function as an organization. Previously, people didn't know. Now we know this is how we do things, how we share everything, we don't care if someone is working on five projects or if someone is working on only one project, when it comes to awards, we would always try to be fair and transparent. Previously, we didn't know. So there was a bit of tension, like, you know, only AI team members are getting highlighted or not. So then I should maybe move to the AI team or not? The other thing I would say is that we sort of now have an unwritten constitution, people know as a group, we generally do these and don’t do these. For instance, everyone knows who is interviewing where, and everyone knows who is preparing for the SAT, GRE, and GMAT, and who is applying. From the very beginning, we were very transparent.
If you're here for six months, we are very happy. Even if you're here for two months, please let us know. So now that people know, no one behaves any differently. So we have a team member who will join Standard Chartered Bank on the first of September. Everyone knows but there will be zero differences in anyone's interaction with him. This is just one example.
So people know, for instance, that we will treat everyone fairly. That in all hands, you can ask any question. Even if you are a data auditor or a call center manager, you can actually challenge our AI lead on whether this particular approach is the best way or not. This has been the case from the beginning. But now, team members who have been with us for a few years, they now know that there's no deviation from this. And the AI lead would not be surprised, they know that this is IM.
To my mind, this is the other thing that you know, we now have a history, and people know, this is how we behave, we value family, we value religion, if someone is from Hindu religion, if someone is from Christian religion, if someone is Muslim, we value everyone. When we design our work, we consider these things.
This is nothing new; it's just that everyone sort of has a clearer understanding now. There's one thing I would say. I often feel that it would have been best if we could have some things more formalized, maybe a grievance policy, maybe a harassment policy. We don't have these written. So what we do is we have Speakout[at]Intelligent Machines, Balun[at]Intelligent Machines. If you write to these email addresses, everyone will receive a copy. So we tried to ensure an arrangement where if anyone is feeling harassed or anything, you can write to us anonymously or with your name. And it's not just for team members; even our suppliers, former team members, anyone can write.
But ideally, what I would have loved is what I've seen at BAT. A clear guideline for what will happen if a request or a complaint comes in. Who will investigate? How long does that take? What will be the repercussions? If you have that, it's another assurance. You know what, there's this sort of safety mechanism. If something happens, I know that five team members would come in, and there'll be at least one woman, at least one from another religion, at least one from another location, and this is how they would, for that time being, keep their account suspended, and this would be the process. I would love to have this here. But the hard fact is, with the current budget we have right now, we need to focus more on continuous survival, immediate survival, and inshallah, maybe in a year or so, we will start working on this.
Right now, in terms of organization, we don't have the staff manual or the guidelines. At BAT, we had guiding principles, and I would love to have that. These are not firm rigid rules, but they are guidelines. Everything now works in a very transparent manner. Everyone is empowered. So anyone can make a decision, but it would have been useful to have guiding principles.
Now, I want to ask you a couple of questions about IM today and the future of IM. What is Intelligent Machines today? Can you tell us about the products and services that you offer? For someone who is not quite familiar with the work you do, how would you describe Intelligent Machines to them and tell them about your products and services?
First, I would like to start by responding to your question about what IM is today and what IM will be in the future. I would request you to imagine it like this: IM today is like a six-month-old baby or a two-year-old toddler. The IM of the future would be like a person in 40 or 60 years, completely different. Right now, we are only an experiment. We are a very important experiment. We are a very small, somewhat successful experiment. But we're still just an experiment. We haven't proven that we can be a thriving business. We haven't yet established that people can come en masse and take our service.
Now, we believe that's not because of the business model, but more because of the timing for consumers in the market space. As of now, AI is a bit early, especially in B2B.
You also requested to know about our products and services. Our products are simply the capabilities we offer. Our product is our capability, nothing else. Please bear with me. So when you buy toothpaste, a keyboard, a pair of glasses, a mouse, or a business card, all of these serve important purposes.
For us, the way we understand it now, this is our current understanding, our current realization: AI is in its very early stage. So, at this stage, it's highly unlikely that you would find many mature products in any industry. You have a higher likelihood of finding a good team than finding a ready product for your needs. What we request from our customers is to think and consider the following: please imagine how it was 40 years ago, 50 years ago, when Microsoft only started, and other software development was just beginning. We didn't have tens or hundreds of SaaS software or ERP software, or even web development software. So, you needed to work with a team who would help build an accounting software for you, who would help build a printing solution for you. Now, the software industry is, mashallah, very mature. You have long portfolios of many teams from where you can buy.
So, again, imagine 2000 years back from now, maybe you didn't have many different styles of dresses to choose from. Today, you can go to a shopping outlet or a clothes store and compare styles, fabrics, and prices that suit your needs. But back then, maybe your best option was to work with a tailor, and that tailor would create a customized solution for you. AI is just like that. So, we request your readers or potential clients to consider this and take it into cognition. Right now, please partner with us for our capability, as we bring in the power of AI and data science to help you solve your highest-value problems. That's it. What are your highest-value goals? Is it optimizing your current operations? Is it reaching new customers? Of course, AI can help. Do you want to build new products? Of course, AI can help.
Let me just give you one specific example. The GPH Ispat Team reached out to us. They mentioned that the way they currently work is by buying containers and containers of steel in auctions in international markets. So, let's say they buy five containers or 20 containers of steel, and these containers contain various materials, such as sheets, bones, utensils, and more. The issue arises when they have specific requirements from different customers for different types of products. To tackle this, they have physical labs with lab operators working in shifts. These operators take stock count of the items they have purchased in auctions and then attempt to sort them out. They think, "Okay, we have these elements. If we melt this and then melt this other sheet part, maybe we can make this rod." However, this process takes time and the resulting composition is often far from optimal.
What AI would do is perform tens of thousands of computations in minutes, significantly reducing the processing time from months to hours. It would provide the optimum recipe, something that humans would not be able to achieve. To be fair, the manual testing in the lab, creating small pieces, breaking them, melting them, and testing their strength, followed by computer simulations on regular computers (not supercomputers), was like comparing an apple to another apple. But with our product, our offering, we provide the capability and strength of AI. Let's take an example from the early 1900s when people didn't know how to utilize the power of electricity in their factories and operations. They needed experts. Now, we are the humble experts in our field.
I recently had a short visit to the USA, where 62 AI experts were invited to a gathering. There, I had a shocking revelation that the situation is the same all over the world. It is a supply-constrained market globally. So, my dream is to become the go-to team worldwide, if possible, by managing resources effectively. I hope that if I had the capability, we would have already developed 2000 highly capable AI engineers from Bangladesh. I believe Bangladesh can produce 2000 AI engineers without any trouble. Beyond that number, I think we will have to work harder. However, up to 2000 AI engineers, I believe we can create that number with graduates from BUET, IUT, SAST, and other universities.
To achieve this, we need to scale our learning. We have identified seven key learnings. One of them is providing large real-life data to students. Currently, students complete courses online or in universities, but they don't get the opportunity to work with actual data. Imagine teaching someone to become a cricketer only through classroom lessons, compared to someone who actually practices with hundreds of balls in the nets or plays cricket games. The capabilities of the two individuals would be completely different. Hence, we need to provide live data as a learning resource.
Another important aspect is providing feedback. Students produce AI models, and we apply those in the market, giving them instant feedback. That's why our team members become production-ready in just two months, and some even in one month, and our internship model works well. This is possible because of the unique positions we have. We provide live data, and our clients provide instant feedback from the market, allowing our people to learn and grow. It's like having a chef's student bake a cake, and the client provides feedback on whether it's good or not. This helps them become better chefs. Additionally, we have mentors like Rashed, who guide you when we face challenges. This fast tracks our learning and growth significantly.
Anyway, our goal is to become that global go-to partner for AI. In our calculation, our clients are gaining 32 taka for each one taka spent on an AI project with us within the first six months. The capability is immense. There is no way people will stay away from adopting AI.
One thing that deeply worries me is that we are witnessing a different form of inequality developing right in front of our eyes. The adoption of AI is disproportionately concentrated in only a few markets in the USA and Europe. As a result, it is not being adopted on a mass level. You can quote me on this, unfortunately, we are not witnessing the large-scale explosion of AI adoption that we should. What is happening is a new form of inequality in terms of capability, where AI becomes a currency. I worry that we will have new "haves" and "have nots," AI "haves" and AI "have nots." As a result, a global company with greater capability will bring in $32, while a local company brings in only $1. The global company will simply outcompete others in the market. So, I fear we might see a new form of colonization if we do not achieve some form of equality and widespread AI adoption across the globe. I don't know.
Maybe one aspect is that it's more to do with behavior and nothing to do with technology or technological capability. The decision-makers are shying away from AI adoption, primarily for two reasons, I would say. One, not realizing the significance, not realizing the loss they're incurring for not taking AI. And the other is they have a fear of embarrassment. I worry that you know, they feel like if I take an AI project, would I have the capability to manage it? What if I can't and then, you know, I embarrass myself, putting my career at risk, and the way the current promotions and everything work, the way corporate promotion and reward mechanism work, it possibly discourages people to take risks, even when the risk could turn out to be good for the organization. They generally want to play safe, and they don't take the risk, which would have been in the better interest of the organization. But because they put their own career aspirations ahead of the organization's goals and interests, unfortunately, many of them shy away.
I want to share that people have the capability.
I hope that we will be able to spread this awareness that you have the capability. You have managed previous technology projects well, you have managed project management, you have managed the internet, you have managed e-commerce, and you have been able to successfully manage previous technologies for your organizations. You have the intellectual capability, managerial, and professional capability to manage AI projects as well. The way I want to see us and more teams from Bangladesh and this region, I pray that we become the global AI center.
That's great! One question about IM: Can you talk a little bit about how big the company is today in terms of the number of people who work there, the number of clients you work with, or any other metrics?
We are still very small. As I said, we are still at an experimental level.
We have 38 team members now. We have 10 clients, coincidentally, 50% of them are in Bangladesh and the other 50% are abroad.
Our clients in Bangladesh are bKash, Unilever, BAT, IDLC Finance, and Prime Bank. Our clients abroad are Telenor, Wedo (a Qatar-based telco), Wave Money, BHP, and Cathay Bank in the USA.
We now have $50,000 in monthly MRR. So, we are still pretty small.
But I believe that if we start activating the market in the right way, if we can communicate effectively if we can make more people aware that AI is so substantially powerful and can be substantially beneficial for their organizations, and if we can give them the confidence that they can manage a project and its risks, then what we have is peanuts. I mean, the whole world is out there. They just are not aware of it now.
How does your international operation work in terms of working with international clients?
Like I said before, all of our clients came to us on their own. All of these international clients have heard about us from someone. After due diligence and interviews, they are now working with us.
With your publication, if you feel it is appropriate, I would like to request that you include this message. If I had the opportunity right now to make a request to our government, I would say this:
I know for a fact that our government is now looking for ways to reasonably support the industry so that we can increase our export earnings. I know that our government has a defined target of $5 billion in export earnings by a certain timeframe.
We have five specific requests, which we believe can not only give our country $5 billion in export earnings later but also earn us 1000 crore taka in revenue right now. This is based on our calculations. Take a look at the way garments work: buyers come to Bangladesh to meet with teams, see the fabrics, and place orders if they are convinced.
I would request our government to create a program where they would go to Fortune 500 and Fortune 100 companies and invite the heads of marketing, finance, and IT to come to Bangladesh. They should proactively tell these executives that they will experience a chaotic airport and messy traffic jams, but to bear with it because it is only a one-time thing. After that, they can meet with the technology teams here, talk to them, and see their developments. Once they see this, they will be convinced.
Right now, Nike and other organizations do not feel convinced or comfortable giving work orders to Bangladesh. Two days ago, I received a request from a US team. They said that Nike is giving this work order to India and the Indian team will give it to us. And they clearly stated that we will be billed $100, but the Indian team will bill $600. Just imagine the loss we are incurring.
We need to convince the buyers that Bangladesh has capable teams. I know my industry colleagues at BASIS are requesting the government to reduce VAT and so on, but I believe that these scaffoldings are not what we need. Rather, we need to go to the source and convince the customers to give us work. If we can do that and get work, it will solve many of our challenges. After all, if the government reduces VAT and income tax, how much extra export would we gain?
My second request is that if our government can do what India did earlier. I heard this one from Babu bhai of BASIS. India went to Citi N.A. and other companies and at the government level, they told them, "Give us work orders, and we will pay for the first six months or so. So you have no risk. And then you see.” And now, Citi N.A. alone has $8 billion in work for Indian IT companies. That's $8 billion in a year from one organization from one country! We cannot have this. If we had the money, we would have made 20 sponsored projects for free. Everyone would see them and then maybe they would be hooked.
The second would have been going to certain companies and telling them, "Give us 20 work orders, and we will sponsor them. You give your work to Bangladeshi companies and see our results."
Third, please make tech ambassadors. Our Senior respected NRBs in the USA and other developed countries who have been there for say decades and working in senior positions in NASA, Amazon, and everywhere, make 10 of them honorary tech ambassadors and make another 10 from local, people who have made great achievements. Then request these 20 to highlight and promote Bangladesh. Because, at the end of the day, you need to convince the buyers.
The next request would have been to publish case studies and advertisements in The Economist, Newsweek, The New York Times, MIT Technology Review, and Harvard Business Review. If the government sponsors and publishes this on a country level, imagine the impact. We need to do this.
We don't have a supercomputer infrastructure. For the high-level work that we do here, we need a supercomputer. We didn't even have a payment mechanism to make the payments for international supercomputers that we rent, and we need to pay a lot more.
Next would be to groom 2,000 AI engineers instead of developing freelancers. Of course, freelancers are important, but for the country, if you want to reach billions of dollars of exports, you'll get that from high-end services. I'm not saying stop those programs, but start this program as well, where we also focus on creating highly skilled talents.
We are in a good position, but if we fail to take advantage, we could become permanently lagging in many of these fields. I would like to communicate a few things. One, people are missing out on AI. Second, the government should pay attention to the development of this field. Marketing works. To that end, I think we need to work to position Bangladesh. Germany as a product has a huge reputation. The same goes for India. People want to work with Indian companies. We need to make Bangladesh as a product a fine and appealing one.
Since we're talking about marketing and communications and how to change the perception, how does your distribution and marketing work now at IM?
Unfortunately, we are very weak in this area. All we do is inconsistent, random, ad hoc communication. The good news is that we have now onboarded a team member, Rajeev. He was actually our second team member, but then he left and worked at other organizations. Rajiv has now come on board. So, inshallah, pretty soon, you will see our new billboard. We will finally start working on a website, and this is how we are thinking. We are thinking that we will have two different communication campaigns: one for the local market and one for the international market.
For the local market, we want the first campaign series to be about how you are missing out on AI. AI gives you a substantial advantage. So, maybe we could do 10-15 small stories where people can see how powerful AI is. The second campaign would be about how you can do AI. The first one is "Get AI and AI does it." The second one is "You can do AI." The third one would be a showcase, an event/exhibition. I want us to rent Drink Gallery or Aliase Fraches and do an exhibition where people can come and see AI examples and then engage with us.
For the international market, I want to complete our website, make a small video, make one PDF, and then have an event maybe in Malaysia or any other country. We would invite team members and then engage with them and publish case studies. I also want to tap into our NRBs, Bangladeshis who are living abroad.
As of now, our marketing is a zero out of five. To be fair to us, we don't have the resources. All the team members are production resources. Everyone, including me, always goes to client meetings and works with teams. We have limited bandwidth now, but we want to change that.
What are the challenges for the company now?
Not exactly challenges per see. These are challenges in terms of things that we have to achieve. Challenges to grow.
One is market activation, and another I would say is cash flow. If we can't activate the market, we would have to operate as a small experiment. And in order to do market activation, we need money and ensure cash flow. We don't have immediate challenges as such. But if we want to go to the next level, we would have to solve these problems.
How do you plan to address these challenges?
We are working on some small communication projects. We are also working on some case studies. We are considering reaching out to some prominent people and asking them to speak about us. If they do so, then after listening to them, people might be inspired. We also want to create some small videos where we want to show the substantial capabilities of AI.
We asked our team members to understand how people communicated when ERP first came out or when electricity first came out. We analyzed these things to understand how to design communication for a new thing. Our challenge is similar. People are not yet seeing AI as a necessity.
People are considering AI as something non-necessary, good to have. They think there are more pressing issues in business. They are considering working on people, process, and those things first. So we are now looking into how new technologies get initial traction. For instance, the refrigerator was not necessary in the past, but it has become necessary over the years.
We are trying to understand how we can take AI to a stage where people would feel that it is a necessity. So our communication objectives are to activate the market and take AI to a position where people consider AI as something necessary. We are understanding the different roles of AI and then exploring ways to make it relevant to people.
This was the last question about IM. Do you want to add anything else that I probably did not ask?
I just got thinking while speaking with you. We are about to see an explosion and copy of a lot of AI things in the coming days.
For students, I have a few suggestions. One, don't merely do online courses. Try to understand the different roles of AI. For instance, in journalism, there are many different roles such as bureau chief, journalist, and reporter.
Say for instance, in healthcare, you have many different roles, such as doctor, nurse, and so on. Similarly, when you are working with AI, there are different roles. For instance, we often need to create a working prototype in a week or so. This is one role. So, understand the role you want to work in and then understand what you would need to learn to work in that role.
How do you operate as the founder and CEO of Intelligent Machines? How do you make decisions? How do you stay productive? It would be interesting if you give us a rundown of your typical day.
My days are really varied. I try to do a few things. I try to read as much as possible. This is a habit I have grown since my childhood. When I was in class four, we used to live in Rampura and my school was in Motijheel. I invented a game to keep myself busy on my way to school and back. We had Murir Tin ( a small school kids transport vehicle) and 6 No buses for commuting in those days.
I would glance at street signs and names of the shops and shut my eyes and try to recall the names I had glanced at. Then I would reopen my eyes and check how many names I could recall correctly.
This game helped me tremendously throughout my life. I didn't know about speed reading at the time, but that is what I was exactly doing at the time.
At the start, I could not recall any names, but gradually I could recall images with a split-second view. Everything is practice. If you practice doing something, your brain will learn accordingly.
What happened was that I could read an entire page in a second. It sounds unbelievable, but I could do it. I remember I read seven books in a day after a bet with a friend.
I no longer have that practice, but I am a voracious reader and I'm reading almost all the time. I try to follow senior people. For instance, I try to take the good parts of Jeff Bezos. I read about these successful people and what they did and how they did it.
Second, I always experiment with new things. For instance, I often change how I operate or make decisions. I was making decisions in a certain way in the past and then after learning a new thing, I change it.
My team members are used to this. They would tell you that Oli bhai always changes his approach.
We have seven leads at IM, including myself. Our tech team has three separate wings from the beginning: AI engineering, Analytics engineering, and product engineering. Our leads are empowered and our operational style is asynchronous.
If something happens, they reach out to me, and if I need them, I reach out to them and call for a meeting.
In terms of my day, it depends on my current focus. For instance, my current focus is communication. Previously, I focused on finance, product, and other areas.
I have focused on different aspects of our operation. In the past, I built client strength, product strength, organizational strength, and financial strength. This is how I see my phases.
Now I've started working on communication. But what happens is that a large chunk of my time goes into stakeholder engagement.
If there is no external demand on my time, I started a 4P approach a few days back where I divide my days across four areas: product, people, strategy, and organization. I allocate 25% of my time to each segment.
But since we are resource constrained, I need to attend to many other things as well. Once we can put together a structure, I hope I will be able to do it properly.
I operate in a flexible manner and run with the priority of the moment.
Now I'm working at the intersection of communication and business development. I'm spending time with some NRB Bangladeshi senior brothers these days. You can call this future business development work.
I usually wake up early. I wake up at around 4 AM and I start working at around 6 AM. By 11:30 AM, I have already worked for 5+ hours. When I feel tired, I take a rest at that very moment. I might have 3-4 hours of sleep at night, but when I feel tired throughout the day, I take a nap or rest for a while. This is super helpful. I get recharged with a short break.
This is a good idea for founders who want to try. As a founder, you work 9-9. You work all the hours, and if you can manage your energy, the rest is alright. I try to manage my energy. I start at 6 AM, I take a short nap after the midday prayer.
However, this is not always possible. Sometimes I have meetings that I can't avoid. In that case, I try to take a nap later on. If I can do it, I'm fully recharged again. One personal tip from me: manage your energy. You can split your sleep if necessary.
Many people might disagree with this, but it has been helpful for me.
And I read. The only way I know is to read all the time. I read all the time. I read a lot on the internet, following the major business publishers and also some of our WhatsApp groups.
But I would also like to mention that you need to make sure you're in control of your attention. If you're constantly checking news and notifications, it is impossible to work.
I don't have any notifications on my phone except for calls, messages, emails, and my calendar. I use other tools such as WhatsApp, and social media when I have time and I can respond to people. I rarely use Facebook. I know my limitations, and I try to operate accordingly. You should control your attention, not notifications.
A couple of books that you would like to recommend to our readers.
Think Again by Adam Grant is a powerful book. I want everyone to read it. It should be taught in schools. It is one of the books I gift frequently.
No Rules Rules by Netflix CEO Reed Hastings is an excellent read.
I'm re-reading CEO Excellence by McKinsey again. It is such a powerful read. I highly recommend it.
There are many books that I enjoy. I love Jeff Bezos's writing. I loved his Invent and Wonder. I also loved Amazon Unbound, both the good sides and the bad sides.
When powerful people want to do good, they can do immense good. The same is true for the opposite. These are powerful learnings.
I have been following the work of some of the most powerful tech founders and CEOs. I have learned that you can never be certain that you will not deviate and become a bad person. That is my number one learning. We all carry the potential to become terrible people.
Number two, we have to be open and flexible. Growing rigid and inflexible is the end of growth. I pray that we do not become such. It is especially important for powerful people to be open to different ideas than their own ideas. If they stop being open, it can bring disaster.
These are some powerful recommendations and reflections. With that, I think this is an excellent place to end this conversation. Thank you so much for taking the time to speak with me.
Cover photo credit: Katherine Taylor