1Mby1M Virtual Accelerator AI Investor Forum: With Naganand Doraswamy, Managing Partner and Founder at Ideaspring Capital (Part 3)

Sramana Mitra: The other point I want to make in this discussion is that AI is not only generative AI. There are many, very powerful problems that are being solved with AI, but not necessarily generative AI.

For example, the domain of medical imaging analysis is not a generative AI problem necessarily. That is an AI problem. So, you don’t necessarily need to converse with your imaging system. You need big data and you need to be able to diagnose things using big data. That’s an AI problem or a big data problem; it’s not a generative AI problem.

Those kinds of companies – Tempest just went public solving that problem. It’s one of the first pure play AI companies to go public. Even though generative AI is dominating the popular consciousness, generative AI is part of the story. It’s not the entire story. That’s something to remember.

The other point that this may be helpful in your thinking about the framework of investing the way I think about it is that we have gone through the whole platform as a service cycle with cloud computing, right?

Salesforce had their CRM product, but very early on, they also brought in this Force.com product. I remember I was invited to the launch of the Force.com accelerator in the mid 2000s and that created the platform on which many vertical cloud companies got built. Veeva was built on that, Actus was built on that; and these are very large companies.

So, if you apply that same architecture and same general model to AI, there are AI PaaS on which you can build vertical products. You don’t have to build the whole stack. You build your own domain specific AI and you build the vertical workflow, and you can have more cost-efficient businesses. These horizontal platforms are attracting incredibly large amounts of money. But you are a relatively small fund. You cannot play that game.

So, you’re going to have to find a strategy of how to invest in companies that can build in a capital efficient manner. In One Million by One Million, we don’t really play in that game either. We are trying to empower as many entrepreneurs as possible as to be able to play. As entrepreneurs, and most of them are bootstrapping first and raising small amounts of money.

There are exceptions – Freshworks raised a very large amount of money that came through our accelerator, but that’s an exception, not the rule. One million companies are not going to raise huge amounts of money. So, I think the Platform as a Service (PaaS) layer of abstraction of infrastructure is really important to empower the large-scale AI application revolution. And it is happening to some extent.

Naganand Doraswamy: You’re right, because there are two, three ways to invest in AI companies. What we have seen happen in Valley also is large amounts are given to some very good entrepreneurs to say, “You guys go figure out, try a few things out, something will work out.” That works in the Valley context where there are large funds and they kind of believe that the entrepreneurs will figure something out. But like you said, we don’t have that luxury.

A lot of what is happening in AI today is still very exploratory. People will try something out and they pick the best of the best and then give you $20 or $30 million to figure out something. Something will happen as you play around; you’ll figure it out.

One of the trends that is happening in India is that there’re a lot of AI services companies coming up. The reason that AI services companies are coming up is they’re all saying, let’s do a few projects and let’s see where there’s a lowest common denominator that is emerging and then go build a product around it, whether it’s a vertical SaaS or whether it’s a platform specific thing.

When you do more services, you start learning from it because it’s such a dynamic field. It’s very hard to say, I’ll build this platform and I’ll build this vertical SaaS. I think that’s another trend we are seeing.

Sramana Mitra: This is a very good point that you brought up because in One Million by One Million, you may have seen my writings on this. We have always supported Bootstrapping Using Services. Well, the very same reason that you just pointed out is that you see the problem recurring in multiple customers and you start solving them with services, but you’re also developing IP and the product and productization in that process. It’s a tried-and-true strategy of building a product by bootstrapping with services. Now, if you transfer that to the AI situation, that is absolutely a great strategy because that gives you access to data.

One of the gaining items in building an AI product is data access. Switch that model to services model, the data is there in the customers shop, and you can access data and train your models with data.

The other thing that I’m seeing is this is something you may want to consider in how you’re investing is that. AI companies in the valley are actually going to market as services companies, not as product companies.

If you look at Palantir, they’ve done it all along.

There’s a company called Machinify that is doing really well and they’ve been doing AI for a while. It’s not a late generative AI kind of company they have been doing. Machinify has been doing AI for now a good five, six years already and are a robust company. But I’ve talked to one of the founders and this company can charge so much more in a services mode. I mean, they’re doing million-dollar contracts in a services mode.

So they use their own product, they do the services on top, and they can charge an enormous amount of money because enterprises are willing to pay if you can show this massive ROI in implementing a solution with AI. People are willing to pay tremendous amounts of money because automation can immediately equate to enormous savings.

Whether you like it or not, AI is replacing people at scale. You can build a solution that is going to replace ten thousand people in an enterprise. Well, people are going to pay millions of dollars for that. Do it in a services mode – product service, technology service, whatever. Implement that, deliver the ROI and develop the case study. You have all the material you need to productize and reference sell those kinds of use cases.

Naganand Doraswamy: We are so early in this entire AI cycle, it’s very hard to conceive and build a product because things are so dynamic. When it’s so dynamic, it becomes very hard to build something that you think will be relevant a year or two years down the road. I think that’s a significant challenge that we are seeing as investors as well, and I’m sure even entrepreneurs who are trying to build products around it face this challenge of figuring out what will be relevant in two years. We have seen in the valley that a lot of AI companies either want to get bought out or they want to sell to somebody because things are moving so fast, it is very hard to keep on being relevant in the industry.

So I think those are some of the key challenges. So, I think it’s a good strategy to go the services way, wade your way through the waters initially, and then understand where things are slowly emerging or converging. That’s one of the good strategies one could use.

This segment is part 3 in the series : 1Mby1M Virtual Accelerator AI Investor Forum: With Naganand Doraswamy, Managing Partner and Founder at Ideaspring Capital
1 2 3 4