Deep Instinct vs Crowdstrike…

We are all concerned about competition and moats in our stocks. Being a large shareholder of Crowdstrike as many of us are, I found the above interview with the chairman of Deep Instinct intriguing and worrying at the same time. He comes from a strong background, former CEO of Palo Alto Networks and prior COO of Zscaler. He referenced prevention as a service, a new category, for his company which provides a monetary guarantee against ransomware of 3 million. They use deep learning which he suggests is superior to what is out there, including his former companies and Crowdstrike. Although they are not coming out to market soon, he said they may move up the timeline.
I’m curious if anyone has any experience with this company. Nvidia is invested in the company.
With the upcoming IPO of Sentinel One, this segment will be interesting to watch and monitor over the next few years.


Interesting. The sort of model he discusses is predictive, not just about recognition. This isn’t entirely dissimilar from the type of AI used to create a self-driving car (in a war-zone?). How much do you need to trust this technology to let it have enough control over your infrastructure to be able to take action, on its own, to prevent an attack before it happens.

I also wonder if the barrier of entry is large. While I am sure the engineering is complex, Crowdstrike is already training machine learning (ML) models, so they have tons of relevant data. This is different to what we’ve been discussing in regards to Upstart, who’s larger and older competitors I’ve been in the business for a long time but haven’t collected the specific type of data they would need to hop in and compete directly with what Upstart is making. Here it seems more like a battle of models. If you have tons and tons of data you just need to put a team together and get them iterating. In fact I wouldn’t be at all surprised if crowd strike had some people working on this. I mean if your whole business is about detecting an attack and then protecting your entire network from it happening again, surely you thought about what it would be like to skip the first step and just protect the whole network from ALL future attacks.

…BUT read my last sentence there. This is the real problem with this sort of application of AI. How do you create something that will recognize all future attacks that humans can think of. And if you could, couldn’t you also create models that could beat the model? It is game theory. I think it will always be an arms race regardless of the technology being used, for the protection or the attack. I can’t imagine crowd strike not getting involved in the race. That would mean they simply stopped caring about innovation. Which, for a tech company, is unthinkable.