Thought Leaders in Cyber Security: Manoj Leelanivas, CEO of Cyphort (Part 3)

Sramana Mitra: Let’s do another couple of examples like this. You said your main differentiation and where you’re innovating the most is in unearthing threats that are unknown to the enterprise. I’m trying to understand what kinds of threats are unknown to the enterprise that your work has helped you figure out.

Manoj Leelanivas: Definitely, the threats unknown is the most important thing. The second part of it, which is probably interesting for you, is that in this modern world, we definitely want to have a solution that is very easy to deploy. We were talking about differentiation, I want to cover how it is a differentiator.

Sramana Mitra: Ease of use, as a differentiator, is not interesting to talk about. We will not extract any insight out of that differentiation. Everybody talks about ease of use. I have no way of knowing whether that’s true or false. Your point that you are able to unearth threats that are unknown to the enterprise is a very interesting angle to develop a story around.

Manoj Leelanivas: The only reason I said it is because if you look at the majority of the security solutions, it takes years to deploy them. We can be deployed in 15 to 30 minutes. Let’s just go back to the primary vector. It’s the combination of behavioural analysis and machine learning we do, which is one of a kind in the industry. You actually can evolve and get better and we evolve with the threats as threats progress. That is the underlying nature of the learning machine.

Sramana Mitra: That would be worth double-clicking down on if you want to talk more about how your machine leaning algorithm works and how it learns. What does it do to be able to figure out new threats? That would be interesting to explore.

Manoj Leelanivas: Like I said, the first element is watching the important input. If you look at the majority’s approach, they look at just one vector. We collect interesting aspects of web, email, inter-employee communication, and we feed all of these indicators to our analytics core. This core analyses these objects using our behavioural analysis engine and then puts it into a machine learning model. At the end of the day, it is a statistical model depending on what particular algorithm you’re using. The more data you have, the more powerful it gets.

We feed our machine learning engines with millions of clean samples and malware samples. We collect not just from our customers and from what’s available from public domain stuff, but we also collect from the wild with our worldwide crawler network. We have a team of research scientists who are second to none in getting those samples.

When you, as a user, are looking at an Adobe PDF file and the PDF is examined by us, we will know exactly that PDF file is malicious or turning to be malicious because we have seen millions and millions of PDF files that are good, and also hundreds of thousands of PDF files that are bad. A new indicator of compromise can be easily siphoned off in that case. That’s how the machine learning model works. The more data you have, the better. Everything is about data these days.

This segment is part 3 in the series : Thought Leaders in Cyber Security: Manoj Leelanivas, CEO of Cyphort
1 2 3 4 5 6