Sramana Mitra: C3.ai is exactly like yours. They have a core platform. They’ve gone after the energy market first. They have relationships with the system integrators. Now they’re broadening. It’s a very similar strategy. There’s another company that I’m tracking that has a similar strategy. They come from an open source route which is H20 AI.
Prashant Kumar: The challenge is that a lot of people are developing point solutions. We focus on what we call a data journey. Your data journey could be data engineering, machine learning, or visualization. C3.ai has an amazing platform. They’ve gone public as well. They’re focused on the machine learning side of things. H20 AI is sitting in a different segment. We don’t play in that space. We integrate with H20 AI. We’re super clear that we’re a data journey platform.
As much as the machine learning is overhyped in the market, teams that deliver on machine learning go for data engineering 80% of the time. If you go to utility and telecom, that whole problem becomes more complicated because the data is constantly changing and evolving. We work in a way where we’re sitting below the likes of Cloudera and Cassandra.
On the northbound of it where before you start doing machine learning, you have to do a lot of data engineering and feature engineering. You can do that within our platform. We are evolving more towards data fabric and data mesh approach.
Sramana Mitra: How many telecom customers do you have, and how many non-telecom customers?
Prashant Kumar: We’ve got roughly 12 to 13 customers. One country will be like a whole customer base. We have projects with Vodafone which spans across the whole of Europe. We’ve got around 10 big telcos as our customers. We are slowly getting into the insurance space.
Sramana Mitra: In these customers, what is the primary value proposition that you bring to the table? Is it the data engineering to feed into machine learning algorithms or is the data organization to store into a data lake?
Prashant Kumar: There’re primarily three use cases. Number one is data injection because of the custom nature of the data. We are streaming data from 11 countries with Vodafone. Europe is complicated as it’s denser. The second problem is we are overlooking old school statistics and scaled big data analytics. 80% to 90% of the operational models in organizations like banks or telcos are still driven by non-machine learning use cases.
Customer 360 is one of our use cases. The third one is where we are predicting customer happiness. The first two value propositions follow the 80/20 rule. 80% comes out of the box and 20% gets configured by a consulting company. The last use case is 50/50. 50% are frameworks and 50% is custom. Every telco is different. This is where we might use H20 AI kind of tools.
Sramana Mitra: This is just my observation. It strikes me that there is room for you to package solutions that are aligned with specific vendors within your ecosystem and bring more complete solutions, and then go to market with some of these vendors. You would have a higher velocity.
Prashant Kumar: Dell sells our product to different markets. We are partnered with Google Cloud. If Google Cloud goes to a mobile operator, we work with them to create the value prop along with Google Cloud. The same thing with AWS. We’re already doing that.
Sramana Mitra: Do you have any other use cases you’d like entrepreneurs to come work on?
Prashant Kumar: We’ve generated too much code around PI systems. Now we’re building data virtualization products. To mobilize the use of machine learning for the end customer, we need to create subject-matter-experts-aligned products. There are too many funded companies that are selling anomaly detection. We can do a lot of anomaly detection, but anomaly detection is still not adopted.
Sramana Mitra: The intersection of vertical and horizontal functional knowledge for any use case is of premium value. That’s where entrepreneurs have a tremendous opportunity.
Prashant Kumar: Absolutely. Everyone tries to build a very scalable platform. The problem with a lot of these intersection use cases is their market size might not be a billion dollars. It’s probably a $10 million business.
Sramana Mitra: Right. Thank you for your time.