My company is Kyndryl, and we work for a UK-based financial institution. That client, the financial institution, has TrendAI Vision One. By using TrendAI Vision One, we are monitoring and doing day-to-day tasks.
In this project, it is related to XDR, but there are many modules. Currently, they are using only HIDS and HIPS. There are many other modules available, but all the modules are based on the license, so they are using only a few of them.
Compared to my previous experience where I worked on some other EDR tools, TrendAI Vision One has many interesting features. There is one module called Cyber Risk Exposure. Inside Cyber Risk Exposure, there are multiple features such as risk overview, exposure overview, and attack overview. In these different overviews, we can easily monitor the overall cyber risk score at an organization level, identify where the loopholes are, and determine where we need to improve security. It monitors the device level, internet-facing assets, accounts, application level, and the cloud. It will show the overall risk based on the different components of the organization. This is a very cool feature for me.
Basically, each endpoint device has an agent called the TrendAI agent, similar to a CrowdStrike agent. The agent is monitoring two things: system events and security events. Based on the events, they are pulling the data at the console for the security team. We monitor if any unusual things happen, and then we have to react. The agent installed on endpoint devices is helping us monitor and do the work.
In a previous company, I used CrowdStrike. Compared to CrowdStrike, TrendAI Vision One is more advanced, I think. I have not used CrowdStrike for more than one year, and maybe they have updated some more features, but I do not remember. However, it has a very good feature, as I mentioned, Cyber Risk Exposure.
Actually, in Cyber Risk, if I want to see the device level and how many devices have some vulnerability, if I click the device, it will show the count. For example, it is showing that two, three, or four devices are detected with this kind of vulnerability. If the devices have Apache Log4j vulnerability or OGNL, then based on the operating system, if the device has Linux, I have to reach the platform team and say "Okay, this system has this kind of vulnerability, and you have to patch the thing" or update the software. From here, I cannot update anything or upgrade the agents. There is some dependency, you could say.
For deployment, I think it is easy and does not require much effort. I have not done the deployment myself, but for some point in time, for a few of the servers, we have done it, and it is easy and does not require much.
For this, it depends on two or three factors. First, we have to confirm why this alert got triggered and what is the IOC. For example, if it is some private IP, then I have to reach out to the different teams. In my case, I have to reach the vulnerability management team because they have Nessus and Qualys tools, which are vulnerability scanner tools. What they mostly do is they try to scan the particular server and devices, targeting the server. When the IP, let us say the Nessus IP, leads to a server, TrendAI Vision One tries to understand "Okay, I think someone is trying to recon this particular server. This is not a usual thing," so they generate the alert. I have to go through each and every alert, and based on whether the IP is private, I have to reach out to the VM team and other teams and try to confirm whether the IP is genuine or from somewhere else. After that, I have to fine-tune inside TrendAI Vision One, and then they will automatically reduce the false positives.
In my case, I can say that earlier we received many alerts related to recon alerts. If I closed and whitelisted two, three, or five IP addresses, the total has been reduced by approximately 40 percent. Earlier, we received more than 400 or 500 false positive alerts, but nowadays we receive hardly 10 or 15 alerts.
My client is not a small bank. I think it is one of the big banks in the UK, but I do not want to tell you the name. It is very big.