1Mby1M Virtual Accelerator Investor Forum: With Suman Talukdar, Founder and Managing Partner at AiSprouts.VC (Part 3)

Sramana Mitra: Let’s dig down to AI and your thesis around AI a little bit. We do a series called Thought Leaders in Artificial Intelligence. It’s a popular series. We have been doing it for a long time. We bring entrepreneurs as well as people in large companies to talk about what they are doing and seeing.

We have a comprehensive thesis on AI. Here are some of the nuggets of what I have observed. I want to hear what you have to say on these conclusions that I have arrived at.

If you were trying to do an AI platform company, it helps if you have done it before. If this is your first time out, is it harder for investors to work with you on a platform company thesis? In that case, a vertical AI, application, domain knowledge, niche, and those kinds of investment thesis work better.

I am a huge fan of the Platform-as-a-Service. One example is CRM which is an application for Salesforce. It’s the platform that gets a broader adoption with an independent software vendor who builds on top of that platform. That model is going to build the biggest companies.

AI has always had this question mark around it. Does it work? Investors have to figure out if it works. In your case, you are saying that you want the product to be working and in the hands of customers using it and getting value out of it. That is not a short time from inception. That early stage of business building and AI is difficult.

If you have credibility from having done it before and you have capital of your own, people will find checks. These are my observations. I would love to hear what you think about the space. 

Suman Talukdar: I generally agree. Entrepreneurship has a lot to do with recognizing if there are exceptions that you have to pay attention to. A lot of it depends on the individual entrepreneurs from that background. There is a third enabler to platforms versus applications.

In platforms, there is a bias towards people who can compete at scale. There are usually products that are not vertically-focused, so it takes more money to sell. You have to raise more capital in this way in the long run. That biases investment towards people that have done it before.

Every so often, there are successful platform companies. I would agree with you that it is helpful. There is a category of enablers in AI. These would be companies developing unique products and services that are sitting on top of the platforms that are not necessarily applications, but they are helping people develop the applications faster.

An example of that is a company called Label Box. They have been good at raising money. Manu Sharma, who is the founder, is a young entrepreneur. It’s his first company out of Stanford. He focused on tagging and what people do after they tag. I invested in that space as well.

Those companies are making the process of using things like Amazon, Google, and developing apps on top of Nvidia easier. That is one example where you don’t have to have this crazy big ambition or capital requirement if you are trying to do a platform.

The third is the vertical application. It’s like the bread-and-butter today of practical or narrow AI. I made a couple of investments in that space. People can take advantage of the fact that there is tons of data. It’s readily available. They can aggregate and store it using tools that were not available before.

With compute like Nvidia and GPU, it has improved 30,000 times in the last five years. The amount to train some basic data sets using standard algorithms has come down quite a bit. What used to take three days, now only takes 60 seconds.

That allows entrepreneurs and their team of two or three people to never buy a server and instead do everything on Amazon to build these focused applications on specific domains. 

Sramana Mitra: The unfair advantage is the domain knowledge. 

Suman Talukdar: Yes. They can tie the domain knowledge with the data and use case. With the application that I am seeing, it’s all about optimizing for very specific use cases. There are a ton of opportunities. 

Sramana Mitra: From our point of view, we want millions of entrepreneurs to be successful. In that verticalized domain-specific AI, we will see a large number of entrepreneurs because people have niche expertise in different areas.

Applying AI to those areas with that domain knowledge is very interesting. The trick is going to be good platform companies enabling that kind of entrepreneurship in a big way. That will be the determining infrastructure that we need at scale to enable this kind of entrepreneurship. 

This segment is part 3 in the series : 1Mby1M Virtual Accelerator Investor Forum: With Suman Talukdar, Founder and Managing Partner at AiSprouts.VC
1 2 3 4 5