Alexander Wong
Co-Founder - DarwinAI

Lessons for Canada’s AI Scale-Ups

Published on


  1. Technology startups should bear in mind that it is never just about the innovation, but rather how that innovation can be applied to industry.
  2. A significant step in scaling up for a startup is to connect with a larger company or a market that has use for their product.
  3. Government support is crucial in order to help drive a culture of technology adoption in Canada.


Government programs that connect Canadian startups with larger corporations and markets need to grow. These programs will greatly facilitate the sharing of technology and expertise, benefitting both the startups with a customer and the larger company with products and solutions to their problems. This extra push could make it significantly easier for startups to scale up.

What has been your experience founding and scaling an AI startup, and what were the main challenges?  

For me, it has been a very enjoyable experience, otherwise I would not be doing it. It is just my way of trying to contribute back to the Canadian economy from an artificial intelligence (AI) perspective, given that is where my expertise lies from a research perspective. One of my main goals was how to bring research from my lab, especially in AI, into the field to benefit industry as well as society.  

There have been a lot of different challenges especially going from a startup and now trying to scale up. For example, one of the most obvious ones has been dealing with the pandemic. With the pandemic, a lot of business interests changed and just getting the team motivated to continue was very important. One of the things that we did and we actually used it as a way to give back to the community, has been the development of the COVID-Net Initiative at DarwinAI, where we took our expertise in AI and created an open-sourced initiative for building deep-learning solutions for helping to screen COVID-19. That is one way that we took the challenge during the pandemic and used it as an opportunity to give back to the community.  

“It is never about just the technology. It is finding a way to actually surface the technology in a form that has the most impact and utility to the customer base.” 

Another key one about growing is, as an academic, I am very familiar with the technology. But at the end of the day, it is never about just the technology. It is finding a way to actually surface the technology in a form that has the most impact and utility to the customer base, or the vertical that you are trying to hit. That has been an important challenge and we received a lot of support from our venture funds that supported us with mentorship to really help us hone in and identify the best ways people can actually leverage our underlying technology. 

Another area that was a challenge as we continued to scale up was building the right team. As a professor, my expertise has been on technology and innovation; it is not about from an organizational, operational day-to-day, or company perspective. We had wonderful people come in like Sheldon Fernandez, with his decades of experience as Chief Executive Officer and Arif Virani with his experience as Chief Operating Officer, and now we continue to face the challenge of how to scale by hiring the right person at the right time, given the stage of our company’s growth. We continue to recruit great talent to cover the different facets that allow us to grow in different areas. 

How can Canada improve its culture of tech adoption? 

Indeed, a major challenge is technology adoption. Within the Canadian ecosystem and even in larger Canadian companies, there is a hesitation to adopt new technology at the same rate as our international counterparts. One of the key things I have seen that have helped a lot is the Canadian government initiative to help with the technology translation. That is something that is a great start but we need to push further. For example, the Supercluster Initiatives are trying to connect small companies, medium-sized companies, large companies, and universities together so they can have the technology transfer flow and benefit from each other. There are also programs such as the National Research Council of Canada Industrial Research Assistance Program (NRC IRAP) and Mitacs programs that help get graduate students into the workplace for technology transfer and innovation. Those are great initiatives. If we can grow those and push those further, it would benefit us significantly. 

What has been your experience with programs like Mitacs, IRAP, and the Superclusters, and how could they be improved? 

We are involved with pretty much all of the above, and those have been very wonderful from my perspective. For example, with Mitacs, we have been able to get really wonderful interns who are really skillful. There are a lot of things that we can offer from a company perspective to help them grow as researchers as well as grow as individuals by joining the Canadian workforce. While they are really great from a fundamental knowledge perspective and bring a lot of new ideas, there are always little quirks with actually translating technology from the lab to industry. There are certain things that will work, and there are certain things in practical, real world situations that might have a certain challenge, be it computing, costs, or outlier situations that academics do not think about. It has been a wonderful experience for us to be able to receive that knowledge and support from an innovation perspective, while giving back to them to help them grow to be good individuals in the workplace.  

“It has been a wonderful experience being connected to other companies who need our products as well as other startups and scale-ups who face similar challenges.” 

It was similar for IRAP and the supercluster, particularly. It has been a wonderful experience being connected to other companies who need our products as well as other startups and scale-ups who face similar challenges who we can actually communicate with. Those initiatives are great and the expansion of these kinds of initiatives would really help support startups such as DarwinAI to go a bit further. There is one area that could really grow, which is in taking more of an international perspective with international exchange, which Mitacs does some of, but also growing up further so that we are working with other larger markets and countries that allow us to share intellectual property (IP), knowledge, and have our individuals intermingle. We have certain great strengths and they have others. The ability to share that amongst different companies of different sizes as well as different countries would be a great way of pushing this. 

What differentiates developing an AI startup from any other business? 

For me there is nothing too fundamentally different. At the end of the day, one realization is: we watch a lot of sci-fi movies, TV shows, and so on and so forth, so a lot of the time people think of AI as this sentient being. They think running an AI company means talking, building, and communicating with this AI being to tell it to help us with our business. That is not how it really works. The way to treat AI is it is another software system in the loop as part of this larger workflow of your system. From that perspective, you would treat it pretty much as any other software. The main difference between an AI company and any other software company is the level of expertise necessary to build this underlying software and the way in which you build it. In your traditional software company, you would think of an algorithm, code it up based on procedures, and then put the different components together. With AI, it is a lot more data driven. Instead of actually building algorithms you are building fundamental blocks of an AI system. For example, we do a lot of deep learning, which means picking different building blocks to put together neural networks and then coding it and building it with data so that it serves a purpose to solve a very complex problem. The skillsets that are involved to do that are actually very different, but at the end of the day, once you build this AI intelligence system software, the way you treat it from an infrastructure perspective is the same as any other software company. 

What advice would you give for successfully scarling a startup? 

One of the key things I would have thought about earlier about is product-market fit. We have this great new technology that is innovative and that no one else has, which is great, but what can it do to serve industry? That is the key thing I would tell myself. Think about that earlier. It is not just that this is a cool, new algorithm and it is going to make magic work for everybody. That is not a practical reality. At the end of the day, even with AI, you need to build it and tailor it in a form that people can see value in. That is the key thing that I would have told myself as well as other AI companies. It is not really about AI first. AI is important, but it is about what you do with it that becomes an enabling factor for industry. Do you allow them to automate certain processes that otherwise would have been left untouched? Do you help improve their overall workflow, and if so, by how much? For example, we do a lot of explainable AI—does it enable you to gain new insights into your business processes for healthcare? Does it give you new insights into diseases? It is very different from the academic mindset, being a professor. If starting over, there are certain things I would do to either get the right contacts and resources to be able to steer from a solution perspective to better cater to these use cases. 

What and who would you pitch in 30 seconds about improving the ability of Canadian companies to scale better and faster? 

I would pitch industrial leaders as well as Prime Ministers, essentially policymakers, as well as those in charge. I would urge policymakers to put greater investment towards helping startups grow, either from a financial perspective or by connecting them with the right mentorship—there have been really wonder incubators around. We have had experience with Accelerator CentreCreative Destruction Lab, and more, but right now they are all very strained in terms of the number of startups they can help. There are only limited resources. To be able to grow those further would help significantly.  

“I would urge policymakers to put greater investment towards helping startups grow.” 

My other pitch is to continue to grow programs like the Superclusters to connect Canadian startups with large Canadian corporations because right now, there is still this big hesitation to get started. Getting started is one of the biggest barriers right now for Canadian companies, so if there are initiatives and funding that allow a large company to work with a startup in a way that allows them to share their technology, something like the Supercluster program but beyond that, that would really help both the startups who are trying to scale up their technology and get the right revenue, as well as the large companies that know they need AI but do not have the internal resources to make that happen. They are always hesitant looking for that catalyst to actually initiate this effort. These kinds of initiatives could really help the startup scale-up ecosystem.  

Alexander Wong
Co-Founder - DarwinAI

Bio: Alexander Wong is the Co-Founder of DarwinAI. He is also currently the Canada Research Chair in Artificial Intelligence and Medical Imaging, Member of the College of the Royal Society of Canada, co-director of the Vision and Image Processing Research Group, and an associate professor in the Department of Systems Design Engineering at the University of Waterloo. He has published over 550 refereed journal and conference papers, as well as patents, in various fields. 


Organization Profile: DarwinAI is an explainable AI company that enables enterprises to build AI they can trust. Their Generative Synthesis technology makes explainability real, allowing developers to understand, interpret, and quantify the inner workings of a deep neural network. The company’s patented explainability technology accelerates advanced deep learning design and unlocks new possibilities for the commercial uses of deep learning.