Reduces MTTR, improves efficiency, and centralizes everythingAI is everywhere, and I feel we need to discover AI for cybersecurity. For the last five years, I myself have been setting up a product and evaluating AI, but I did not do it directly in Splunk because there was some fear about the performance on the platform. That is why we chose to build our own platform based on S3 and AWS to execute and run all the algorithms and send the data back into Splunk. We now have a good base with innovation. We have a better base to directly include machine learning use cases in Enterprise Security, but they need to provide more capability and autonomy to the customers because there is so much to do. When you are lucky enough to have a team of data scientists, they want to have free hands, so a balance has to be found between giving a lot of autonomy and putting enough control so that they do not do something stupid on the platform, which can impact the performance of the standard use cases as well. There is a compromise there. However, AI is growing so fast that you absolutely need to give autonomy to the team to manage and utilize AI. This means providing easy access to a new library to build machine learning. They need to give us some flexibility, and it seems that with the new toolkit, it would be okay. Data scientists love to have freedom. Splunk Enterprise Security provides us with the relevant context to help guide our investigations, but it would be interesting to add even more context, for instance, in order to raise the level of risk. That is something that can be implemented. For instance, when it comes to data loss prevention, it is known that somebody who is about to leave the company is much more likely to take data. If you have this information soon enough, because the person will tell HR in advance, you can increase the level of risk for data loss for that user, but you do not always get to know that in advance. In such cases, some indicators in the behavior of the users who are leaving could be connected to AI. For instance, if you are using natural language programming, you can grab emails with words like farewell. There are plenty of keywords that you can catch that would give you an indication that the person is about to live. When you have internal and external partners, they can leave at any time. You do not always get the information in advance, but there are always farewell parties or goodbye emails. You can use this kind of information to raise their risk on the specific alert. It is efficient in the sense that you decrease your false positive rate. To me, AI is something that can help them accelerate and improve on what they are already doing. Splunk Enterprise Security is very costly. Pricing is probably its weakest spot.
