WEBVTT 00:00:00.000 --> 00:00:07.000 Let's assume an agent is acting like a very stupid person and you ask your agent, 00:00:07.600 --> 00:00:10.000 hey, I need some fresh air, can you help me with that? 00:00:10.700 --> 00:00:18.400 And the agent could just open a window for you, I mean, a real window, or as you stated, 00:00:19.000 --> 00:00:24.100 even though the intention is there, the agent could take a different path to the solution. 00:00:24.900 --> 00:00:31.900 So, if the agent is not limited in options, he could also punch a hole into the wall 00:00:32.200 --> 00:00:34.600 and that would also lead to fresh air. 00:00:48.800 --> 00:00:53.300 Welcome to the KuppingerCole Analysts Chat. I'm your host, my name is Matthias Reinwarth, 00:00:53.300 --> 00:00:56.550 I'm analyst and advisor at KuppingerCole, and 00:00:56.550 --> 00:00:59.800 so is Phillip Messerschmidt. He's a colleague 00:00:59.800 --> 00:01:05.500 of mine in advisory, he's deputy chief of advisory. Hi, Phillip, good to have you. 00:01:06.500 --> 00:01:08.000 Hi, Matthias, thanks for the invitation. 00:01:08.400 --> 00:01:12.900 Great to have you, and this time we really want to have a chat. This is not an interview, 00:01:13.200 --> 00:01:19.900 we want to look back at an event that happened at the time of this recording, exactly, 00:01:19.900 --> 00:01:23.065 more or less exactly four weeks ago. So, it was 00:01:23.065 --> 00:01:26.500 the EIC, the European Identity and Cloud Conference 00:01:26.900 --> 00:01:32.900 in Berlin. We are working for KuppingerCole Analysts, but this is really the flagship 00:01:32.900 --> 00:01:38.000 event when it comes to identity management and everything around that for Europe. 00:01:38.500 --> 00:01:41.429 You've been there, Phillip, I've been there, we've 00:01:41.429 --> 00:01:44.300 worked a lot, we've talked a lot, we have not had 00:01:44.300 --> 00:01:47.900 the chance to watch everything because we had 00:01:47.900 --> 00:01:51.900 five tracks and a lot happened. You've been there, 00:01:52.100 --> 00:01:55.500 Phillip, from your first impression, when you look back at this EIC, 00:01:56.800 --> 00:01:58.889 was it different? Was it bigger? Was it, 00:01:58.889 --> 00:02:01.500 yeah, was it different? Was it a change to others? 00:02:01.800 --> 00:02:04.900 What is your striking memory when you think back of EIC? 00:02:05.200 --> 00:02:09.400 Well, I think it's not that easy to say it was different than the others. I mean, 00:02:10.300 --> 00:02:13.429 it was the same to that extent that we had a 00:02:13.429 --> 00:02:16.700 lot of topics less than a couple of years ago, 00:02:16.800 --> 00:02:19.335 but still enough. The quality of the 00:02:19.335 --> 00:02:23.700 presentations, that's at least what I heard from the audience, 00:02:24.400 --> 00:02:26.971 was, again, very good, a very high level of 00:02:26.971 --> 00:02:29.900 quality. So, in that terms, it was not different. 00:02:30.300 --> 00:02:33.986 It was maybe better, but I would say it was 00:02:33.986 --> 00:02:38.100 the same like in the last years. What changed is 00:02:38.100 --> 00:02:41.547 the topics. So, what we have seen in the past 00:02:41.547 --> 00:02:45.300 is that we usually had a couple different trends. 00:02:45.800 --> 00:02:50.900 This year, it was a little bit different. It focused a lot on AI and agentic AI, 00:02:51.400 --> 00:02:58.400 but there were topics right and left of that and not as prominent as AI and agentic AI, 00:02:58.600 --> 00:03:02.100 but we still have captured a couple trends besides that. 00:03:02.800 --> 00:03:08.100 And AI and agentic AI and non-human identities, we can discuss that term later, 00:03:08.900 --> 00:03:13.600 really overshadowed a lot, but that does not mean that there was not more around it, 00:03:13.600 --> 00:03:18.800 just as you said. So, it was, of course, the elephant in the room to have this agentic AI 00:03:18.800 --> 00:03:22.376 topic there, and I did a lot around that as 00:03:22.376 --> 00:03:26.700 well, workshops and panels, and every second speaker 00:03:26.700 --> 00:03:29.600 at least had a mention of that topic in their 00:03:29.600 --> 00:03:32.500 presentation as well. So, it is a huge thing, 00:03:33.400 --> 00:03:40.000 and maybe we should also talk a bit about that, but we need to make sure that we also 00:03:40.000 --> 00:03:42.602 cover all the other topics that are as 00:03:42.602 --> 00:03:46.300 important, different important, differently important, 00:03:46.400 --> 00:03:51.800 but I think there are more to come. But start with the agentic AI topic. So, 00:03:52.600 --> 00:03:57.098 is it, the last years in EDIC, if I think back, 00:03:57.098 --> 00:04:01.500 it was feeling a bit of more of the same and a 00:04:02.100 --> 00:04:07.300 steady evolution. This time, it felt like disruption, revolution, things changing, 00:04:07.800 --> 00:04:13.800 old paradigms break, and I think that changed the tone. Do you agree? 00:04:14.100 --> 00:04:21.600 Again, to a certain degree. I would say AI and agentic AI are a new challenge for us, 00:04:21.800 --> 00:04:28.900 because when we think about the NHI in general so far, that were like technical identities 00:04:29.500 --> 00:04:35.500 that served a certain purpose, and we were aware of that purpose. We knew that purpose. 00:04:36.000 --> 00:04:38.755 We were, I wouldn't say in control, but we 00:04:38.755 --> 00:04:42.100 would have the chance to be in control of all NHIs. 00:04:42.900 --> 00:04:48.400 With agentic AI, it's a little bit different because it acts autonomously, 00:04:49.300 --> 00:04:58.900 and that makes agentic AI more comparable to a human identity than to a classic NHI, 00:04:58.900 --> 00:05:08.300 like we know it. And the interesting part about agentic AI is, with acting autonomously 00:05:09.100 --> 00:05:14.711 and without, let's say, a framework of ethics, 00:05:14.711 --> 00:05:20.200 the agent can potentially act like an insider 00:05:22.600 --> 00:05:29.400 and can create an insider threat. So, like a human, but without the idea of what is good 00:05:29.400 --> 00:05:33.300 and what is evil. And that's interesting because 00:05:33.300 --> 00:05:37.200 in IAM, we built our frameworks and our controls 00:05:37.200 --> 00:05:44.900 and everything around human identities and non-human identities that served a certain 00:05:44.900 --> 00:05:50.323 purpose. We have no frameworks that can handle 00:05:50.323 --> 00:05:56.100 an autonomous AI without a good ethical framework 00:05:56.100 --> 00:06:00.353 behind that. So, I think the behavior or the 00:06:00.353 --> 00:06:04.800 agency of an agent is really something that we 00:06:05.200 --> 00:06:08.614 really need to wrap our heads around right now. 00:06:08.614 --> 00:06:12.100 This is something where also the industry agreed 00:06:12.100 --> 00:06:14.104 that we are not yet there, that not every 00:06:14.104 --> 00:06:16.500 solution is already on the table when it comes to 00:06:17.300 --> 00:06:20.367 formulating intent as something that is passed 00:06:20.367 --> 00:06:23.500 between different agents or from a person to an 00:06:23.500 --> 00:06:28.100 agent, to another agent, to an application server, to a database, to an MCP server. 00:06:28.500 --> 00:06:31.767 So, that is really something where things change. 00:06:31.767 --> 00:06:35.100 But on the other hand, we are dealing with systems 00:06:35.100 --> 00:06:38.138 or with requirements to systems that we did not 00:06:38.138 --> 00:06:41.500 have before. So, we are talking about machine speed, 00:06:41.700 --> 00:06:44.894 not people speed. I request access, I approve 00:06:44.894 --> 00:06:48.300 access, I assign access. This happens now within 00:06:48.400 --> 00:06:51.949 milliseconds. So, this is something our systems 00:06:51.949 --> 00:06:55.800 were not built for. And the same is true for scale, 00:06:56.100 --> 00:07:02.700 for the sheer velocity, the amount of identities that are around. And then they are also 00:07:03.400 --> 00:07:08.700 agentic, they have agency. They solve their issues and their tasks for themselves, 00:07:09.100 --> 00:07:16.900 just as you said. And they don't care if it's legal or illegal. So, it is ethics because 00:07:16.900 --> 00:07:19.126 they don't care about ethics, they just solve 00:07:19.126 --> 00:07:21.500 the challenge. And this is something that we are 00:07:21.900 --> 00:07:24.376 currently dealing with. And I think the industry 00:07:24.376 --> 00:07:26.800 is doing a great job in getting closer to that, 00:07:26.800 --> 00:07:29.765 but we're not yet there. So, I think this is 00:07:29.765 --> 00:07:33.000 really something that really changed. So, we are 00:07:33.000 --> 00:07:39.600 breaking paradigms that worked for decades. Yeah, absolutely. And we can see that with the 00:07:39.600 --> 00:07:43.397 trends that we have discovered at EAC. So, as 00:07:43.397 --> 00:07:47.700 said, it's not all about agentic AI, but we can use 00:07:48.500 --> 00:07:51.836 some of the known technologies and some of the 00:07:51.836 --> 00:07:55.100 known methods that we already have in IAM for 00:07:55.100 --> 00:08:00.800 at least solving parts of the challenges with agentic AI. So, I'm talking about dynamic 00:08:00.800 --> 00:08:08.200 authorization, for example. We need to work with the agents that come at machine speed, 00:08:08.400 --> 00:08:13.500 making decisions at machine speed. And a challenge that we currently have is that 00:08:13.500 --> 00:08:21.300 the current authorization models are built to work with static entitlements, with static 00:08:21.300 --> 00:08:29.600 situations. And agents at agentic AI, they are no static challenge. They act much faster, 00:08:30.700 --> 00:08:37.300 much more unpredictable than any human. As I mentioned earlier, they are autonomously 00:08:37.800 --> 00:08:42.475 acting identities without a sense of good or 00:08:42.475 --> 00:08:48.000 bad. And that makes them a threat if we are not able 00:08:48.000 --> 00:08:54.500 to give them the guardrails. And dynamic authorization can make that happen. So, 00:08:55.200 --> 00:09:02.500 one very interesting presentation from one of the vendors and one of the statements that 00:09:02.500 --> 00:09:10.600 the person made was that we need to protect the data. It doesn't help us just to identify 00:09:10.600 --> 00:09:14.037 and discover the agents, but we need to protect 00:09:14.037 --> 00:09:17.400 the data. So, instead of discovering thousands 00:09:17.400 --> 00:09:24.400 or ten thousands of agents, it's much more valuable to protect all the data points or 00:09:24.400 --> 00:09:29.400 the applications and put up the policy enforcement points in front of our data, 00:09:29.900 --> 00:09:36.800 ensuring that the agent needs to identify himself and the intention that the agent has. 00:09:38.100 --> 00:09:41.250 And that is, I think, an important point here 00:09:41.250 --> 00:09:44.400 when it comes to the trends that I mentioned. 00:09:44.400 --> 00:09:51.200 So, we can see agentic AI requires the other IAM topics. 00:09:51.500 --> 00:09:56.900 I fully agree. Protecting data is, when it comes to security and also to governance, 00:09:57.000 --> 00:09:59.550 the first starting point. I would not say that 00:09:59.550 --> 00:10:02.100 it does not make sense to identify the agents, 00:10:02.200 --> 00:10:08.700 but in the end, when this thing, this agent, that process, that autonomously acting 00:10:09.700 --> 00:10:13.600 instance of software, when it makes a decision, when it transfers money, 00:10:14.400 --> 00:10:21.600 you want to know what agent was it and on whose behalf it acted or who is liable for that, 00:10:21.700 --> 00:10:29.800 who's responsible, who's the owner. But that in a pyramid of controls comes second place, 00:10:29.800 --> 00:10:32.225 because you don't know which agent knocks on the 00:10:32.225 --> 00:10:34.600 door the next time. And maybe it's a known one, 00:10:34.600 --> 00:10:37.750 unknown one, first protecting the data and then 00:10:37.750 --> 00:10:40.900 later understanding what happened right now and 00:10:40.900 --> 00:10:47.700 who's responsible and why was it correct or not. Both dimensions are absolutely valuable 00:10:48.100 --> 00:10:50.885 and we need to find the proper way to deal 00:10:50.885 --> 00:10:54.200 with that. But protecting data in the first place, 00:10:54.300 --> 00:10:58.265 I would fully agree. On the other hand, what 00:10:58.265 --> 00:11:02.500 we are still struck with, and you said they are 00:11:03.200 --> 00:11:06.280 like people, is that you can give them the 00:11:06.280 --> 00:11:09.800 same task and maybe next time they will choose a 00:11:09.800 --> 00:11:12.850 different approach to solve the same task, which 00:11:12.850 --> 00:11:15.900 is a good thing for a person and a regular thing 00:11:15.900 --> 00:11:18.151 for a person. This is what we're used to. Some 00:11:18.151 --> 00:11:20.500 ways are better than the other and some are more 00:11:20.500 --> 00:11:23.352 effective than the other and these things 00:11:23.352 --> 00:11:26.900 learn. But they are non-deterministic and that is a 00:11:27.000 --> 00:11:35.300 challenge that we as IT people who think in algorithms are not yet used to. And this is 00:11:35.300 --> 00:11:38.950 something that we also need to translate into 00:11:38.950 --> 00:11:42.600 the proper controls. Jonathan, our colleague, 00:11:43.000 --> 00:11:49.900 mentioned the idea of having something like ATDR, which is of course a version of ITDR, 00:11:49.900 --> 00:11:53.658 Identity Threat Detection and Response, apply 00:11:53.658 --> 00:11:57.500 this for agents. It's not just the same set of 00:11:57.500 --> 00:12:03.700 capabilities, but it's machine speed, machine velocity, machine scale and being able to 00:12:04.300 --> 00:12:07.200 understand what the intent was. And I think this 00:12:07.200 --> 00:12:10.100 is where things get spicy. Do you agree as well? 00:12:10.200 --> 00:12:14.853 I agree, but I think the way to get there is not 00:12:14.853 --> 00:12:19.700 as easy as it sounds. Let's take a simple example. 00:12:20.600 --> 00:12:24.150 If you tell your agent, let's take a real-life 00:12:24.150 --> 00:12:27.700 example where obviously no agent can help you, 00:12:28.400 --> 00:12:36.100 but let's assume an agent is acting like a very stupid person and you ask your agent, 00:12:36.500 --> 00:12:39.900 hey, I need some fresh air. Can you help me with 00:12:39.900 --> 00:12:43.300 that? And the agent could just open a window for 00:12:43.400 --> 00:12:52.000 you, a real window. Or, as you stated, even though the intention is there, the agent could 00:12:52.000 --> 00:12:56.263 take a different path to the solution. So, if 00:12:56.263 --> 00:13:01.000 the agent is not limited in options, he could also 00:13:01.000 --> 00:13:07.133 punch a hole into the wall and that would also 00:13:07.133 --> 00:13:13.400 lead to fresh air. That was your intention. You 00:13:13.400 --> 00:13:20.400 not your intention in the end. And that is something that we see in technology as well, 00:13:20.500 --> 00:13:23.800 but it's a little bit more abstract and harder to understand for people. 00:13:25.300 --> 00:13:29.311 So, with agents analyzing and covering 00:13:29.311 --> 00:13:34.800 agents' intent, that would mean that we need another 00:13:34.800 --> 00:13:42.500 agent ensuring that the first agent is not punching a hole in the wall. Both agents need 00:13:42.500 --> 00:13:47.950 to learn that. So, the monitoring agent needs to 00:13:47.950 --> 00:13:53.400 have that understanding of what is good and what 00:13:53.400 --> 00:13:56.044 is bad. And that is something that the agents 00:13:56.044 --> 00:13:59.100 need to learn before they can start their monitoring 00:13:59.200 --> 00:14:08.600 work in the end. And interrupting the first agent basically before the mistake happens. 00:14:09.200 --> 00:14:14.800 So, that is something that they need to learn and that we need to teach them 00:14:14.800 --> 00:14:17.697 before we can apply that. But in the end, I 00:14:17.697 --> 00:14:21.200 think the statement itself is right. When we want to 00:14:21.200 --> 00:14:24.518 use the advantages of the agents at machine 00:14:24.518 --> 00:14:28.300 speed and at machine scale, the human in the loop 00:14:28.300 --> 00:14:31.366 cannot be the solution, at least not at scale 00:14:31.366 --> 00:14:34.500 and not at the speed. It can be a solution for 00:14:34.500 --> 00:14:38.099 the learning and the learning algorithm for 00:14:38.099 --> 00:14:42.200 the, let's say, monitoring and governance agents, 00:14:42.800 --> 00:14:49.900 but it can't be the solution to ten thousands of agents running around at machine speed. 00:14:50.000 --> 00:14:57.100 And that is basically something that Jonathan stated, that we can expect ten thousands of 00:14:59.400 --> 00:15:04.500 agents compared to one human as a ratio in around 12 months. So, 00:15:05.300 --> 00:15:08.900 let's see if that really happens. What do you think about that? 00:15:09.600 --> 00:15:12.816 One thing is clear. No matter how we achieve it, 00:15:12.816 --> 00:15:16.100 we need to have something that we did not have in 00:15:16.200 --> 00:15:18.515 traditional identity and access management. 00:15:18.515 --> 00:15:21.100 Usually, we will need to have behavior analytics 00:15:21.100 --> 00:15:29.600 for agents at runtime. And as you said, the human in the loop is not really an option. 00:15:29.800 --> 00:15:37.100 It might help in some edge cases, but that's it because speed, velocity does not scale 00:15:38.000 --> 00:15:41.586 with people. Changing access management, access 00:15:41.586 --> 00:15:45.400 governance for agents by using behavior analytics, 00:15:45.500 --> 00:15:47.951 I think that is a starting point. How we do 00:15:47.951 --> 00:15:50.800 that, is it another agent that controls the agent, 00:15:50.800 --> 00:15:57.500 but who controls that agent? That will be the interesting part. But when something happens 00:15:57.900 --> 00:16:06.900 with its own intent, with its own plan, its own agency, and its own strategy, we need to 00:16:06.900 --> 00:16:09.819 monitor the strategy and how it's executed. And 00:16:09.819 --> 00:16:12.800 that is something that we have not been yet used 00:16:12.800 --> 00:16:17.000 to do that. But we need to get there and we need 00:16:17.000 --> 00:16:21.200 to get there fast. This is really something that 00:16:21.400 --> 00:16:27.800 we need to do. And just one final sentence. When Martin suggested to do a pre-conference 00:16:27.800 --> 00:16:37.300 workshop on the changes of AI to IAM a week before we started EIC, the team that actually 00:16:37.300 --> 00:16:45.000 prepared for that sent in questions and areas where AI changes IAM. And within two days, 00:16:45.200 --> 00:16:50.900 we had 15 changes and all were valid. And we thought we stick with them. And then we added 00:16:50.900 --> 00:16:54.333 up with 15 more. And I think we're still not 00:16:54.333 --> 00:16:58.000 yet complete. If somebody's interested in that, 00:16:58.400 --> 00:17:00.421 please look at the slide deck and it's available 00:17:00.421 --> 00:17:02.400 and the recording is available for those who've 00:17:02.500 --> 00:17:06.061 been there. So there is a lot that's changing. 00:17:06.061 --> 00:17:09.700 But actually, we wanted to look into the topics 00:17:09.700 --> 00:17:13.211 that are not AI and that are not agentic. And 00:17:13.211 --> 00:17:16.800 so the question is, and we've already had this 00:17:16.800 --> 00:17:19.443 overshadowing even from what we've been 00:17:19.443 --> 00:17:22.900 talking about. So stop it here. This won't go away, 00:17:23.100 --> 00:17:25.837 this topic. We will have that for years. That 00:17:25.837 --> 00:17:29.000 will change a lot in identity and access management. 00:17:29.700 --> 00:17:34.000 But if you look back at EIC, what were the other aspects? You've mentioned dynamic 00:17:34.000 --> 00:17:39.002 authorization. What else happened? Well, we have 00:17:39.002 --> 00:17:43.900 the always coming back topic of UID wallets and 00:17:43.900 --> 00:17:51.400 the success factors of this topic. We have a sovereignty, data sovereignty. That is also 00:17:51.400 --> 00:17:59.500 the thing that becomes more important, especially now when we think about the geopolitical 00:18:00.500 --> 00:18:05.678 world and how it is at the moment. We still 00:18:05.678 --> 00:18:11.700 need to think about cryptographic agility and what 00:18:11.900 --> 00:18:18.700 happens when the Q day arrives. So that is something that we need to prepare for. 00:18:18.800 --> 00:18:25.200 And then we have the identity fabric and the strategic view on all of the topics and how 00:18:25.200 --> 00:18:32.100 we can handle that based on structured frameworks. I think that's also important. 00:18:33.000 --> 00:18:38.200 Yeah. And most of the other trends we have already mentioned with our AI talk earlier. 00:18:38.700 --> 00:18:42.489 So these are the four that we have left, I 00:18:42.489 --> 00:18:47.000 would say. So which one do you want to start with? 00:18:48.800 --> 00:18:52.677 I would add one because we will go back to 00:18:52.677 --> 00:18:57.200 yours. But the topics that you've just mentioned, 00:18:58.000 --> 00:19:02.100 they are real life topics, but they are challenging for many organizations 00:19:02.300 --> 00:19:05.202 when it comes to planning for the future. So 00:19:05.202 --> 00:19:08.500 they are somewhat future looking sovereignty. It's 00:19:08.500 --> 00:19:11.293 difficult because systems being available, crypto 00:19:11.293 --> 00:19:14.200 agility. Yeah. What do I do? Exchanging algorithms. 00:19:15.100 --> 00:19:18.502 So these are new topics. We had a set of great 00:19:18.502 --> 00:19:22.200 presentations and panels that covered the topic of 00:19:22.200 --> 00:19:25.410 how do I make my own now existing identity and 00:19:25.410 --> 00:19:28.900 access management program slash project a success. 00:19:29.400 --> 00:19:33.948 So just back to basics, back to the real life 00:19:33.948 --> 00:19:38.900 key performance indicators, the best practices of 00:19:38.900 --> 00:19:42.540 how can I achieve the best for my organization 00:19:42.540 --> 00:19:46.100 when it comes to very traditional human-based 00:19:46.100 --> 00:19:48.374 identity and access management. There were a lot 00:19:48.374 --> 00:19:50.600 of talks about that, and this is still a topic. 00:19:51.300 --> 00:19:55.450 And the EIC also needs to make the full balance 00:19:55.450 --> 00:19:59.600 between bread and butter topics and these shiny 00:19:59.600 --> 00:20:02.834 new AI topics that will influence our world for 00:20:02.834 --> 00:20:06.000 the next 15 years. And I think that is also an 00:20:06.200 --> 00:20:10.600 important part to have the practitioners exchanging with each other. How do you 00:20:10.600 --> 00:20:17.600 choose the right method to roll out the next authentication factor? And on the other hand, 00:20:17.900 --> 00:20:26.200 the really challenging new topics of AI. And if I go back to your question, I think 00:20:27.400 --> 00:20:30.400 the question of sovereignty is really something 00:20:30.400 --> 00:20:33.400 that is around right now. So when I say we just 00:20:33.400 --> 00:20:41.200 recorded this four weeks after EIC and between EIC and today, there is the switching of AI 00:20:41.200 --> 00:20:49.900 capabilities within Anthropic based on a governmental decision by the US government. 00:20:50.700 --> 00:20:53.650 And that is something that influences a lot of 00:20:53.650 --> 00:20:56.600 organizations that are actually already trying 00:20:56.600 --> 00:21:01.600 to use these capabilities. So there are, without any political interpretation, 00:21:02.400 --> 00:21:05.640 we need to deal with changes that we did not 00:21:05.640 --> 00:21:09.100 have to do that much before. And I think that's 00:21:09.100 --> 00:21:11.519 a big challenge. And we can look at that from 00:21:11.519 --> 00:21:14.100 various aspects when it comes to data residency, 00:21:14.600 --> 00:21:17.784 when it comes to who processes the data and who 00:21:17.784 --> 00:21:20.900 has access. I think that is a hugely important 00:21:20.900 --> 00:21:24.030 topic and it's really on the mind, especially 00:21:24.030 --> 00:21:27.300 of CISOs and of CEOs that want to protect their 00:21:27.300 --> 00:21:32.600 intellectual property and their employee data and their customer data. Don't you agree? 00:21:33.000 --> 00:21:37.453 I agree 100%. We see that currently a lot when 00:21:37.453 --> 00:21:42.100 it comes to data residency. So especially highly 00:21:42.100 --> 00:21:46.354 regulated organizations, they don't want their 00:21:46.354 --> 00:21:50.700 data to go outside of their control, outside of 00:21:50.700 --> 00:21:59.400 their reach, outside of, in that case, of very often Europe. And they don't really trust, 00:21:59.800 --> 00:22:05.100 I mean, the hyperscalers in general, they still trust. But especially when we talk about 00:22:05.100 --> 00:22:09.213 software as a service, the trust in vendors 00:22:09.213 --> 00:22:13.900 that deliver from outside of Europe to a European 00:22:13.900 --> 00:22:22.300 organization, that trust feels lower than one or two years before. So data sovereignty, 00:22:22.900 --> 00:22:29.600 or in general sovereignty, is a topic that is really important. 00:22:29.800 --> 00:22:33.486 Yeah, I absolutely agree. And sovereignty and 00:22:33.486 --> 00:22:37.500 also creating the right fabric for your services, 00:22:37.700 --> 00:22:40.099 not necessarily an identity fabric, parts 00:22:40.099 --> 00:22:43.200 thereof, but also to create a fabric of services that 00:22:43.200 --> 00:22:46.514 isolates this data and these services that are 00:22:46.514 --> 00:22:49.900 most important for you, while still integrating 00:22:49.900 --> 00:22:55.200 with other ecosystems. I think that is also something that we need to deal with, which I 00:22:55.200 --> 00:22:57.597 tried to be a good segue, but it did not work 00:22:57.597 --> 00:23:00.100 that well towards standards. Because of course, 00:23:01.100 --> 00:23:03.484 we need standards if you want 00:23:03.484 --> 00:23:08.500 interoperability. And the more building blocks we have in our 00:23:08.500 --> 00:23:11.717 fabric, cybersecurity fabric, AI security fabric, 00:23:11.717 --> 00:23:15.000 identity fabric, the more components are in there, 00:23:15.300 --> 00:23:22.600 the more we rely on proper integration via standards, via APIs. And this is something that 00:23:22.600 --> 00:23:26.809 I, of course, want to mention as well, because 00:23:26.809 --> 00:23:31.200 a few weeks slash months ago, you and I welcomed 00:23:31.200 --> 00:23:38.800 David Brossard of CEN within this podcast, where we talked about their establishment of 00:23:38.800 --> 00:23:49.200 standards and profiles to allow for proper integration. And the CEN organization won 00:23:49.200 --> 00:23:52.695 an EIC award for their really impressive work 00:23:52.695 --> 00:23:56.500 regarding the standards. If I remember correctly, 00:23:56.500 --> 00:24:03.100 Philipp, you did actually the Laudatio for the EIC award. So what was it that they were 00:24:03.500 --> 00:24:09.200 awarded with this award for? What was their actual achievement that you highlighted? 00:24:09.500 --> 00:24:11.704 I think the most important thing is the 00:24:11.704 --> 00:24:14.700 dynamic authorization part. I mean, they started that 00:24:14.700 --> 00:24:18.150 work group a couple of years ago, when agentic 00:24:18.150 --> 00:24:21.900 AI and the challenges that now arise were not even 00:24:21.900 --> 00:24:31.200 a topic. So the original idea was just to standardize dynamic authorization beyond 00:24:31.800 --> 00:24:35.743 the actual solution. So making all the vendors 00:24:35.743 --> 00:24:39.600 being able to work with each other so that we 00:24:39.600 --> 00:24:43.713 don't see the same issues like with the IGA 00:24:43.713 --> 00:24:48.400 vendors, that every vendor has its own connectors 00:24:48.400 --> 00:24:53.100 to the target systems. And with dynamic authorization, and then especially, we now 00:24:53.100 --> 00:25:01.500 have a standard that supports that interoperability. And the interesting piece now is with 00:25:01.500 --> 00:25:05.795 agentic AI coming up, this standard is also able 00:25:05.795 --> 00:25:10.000 to support agentic AI and the MCP communication 00:25:10.300 --> 00:25:12.402 in the background. So even though it started 00:25:12.402 --> 00:25:14.600 a couple of years ago, it is solving a problem 00:25:14.600 --> 00:25:18.111 that we were simply not aware of at that time 00:25:18.111 --> 00:25:21.700 when it started. But now it serves the purpose 00:25:21.700 --> 00:25:24.211 very well. Right. So they started the work 00:25:24.211 --> 00:25:27.200 right in time to solve issues they didn't know of. 00:25:27.400 --> 00:25:36.700 This is a good starting point. And maybe if you pick one more topic, and I think that is, 00:25:37.300 --> 00:25:40.762 I want to end with a quote that Martin said in 00:25:40.762 --> 00:25:44.300 his closing keynote. And this is something that 00:25:44.300 --> 00:25:46.918 we both agreed before we started this recording. 00:25:46.918 --> 00:25:49.700 This is a striking sentence, and it's unfortunately 00:25:49.700 --> 00:25:56.200 very, very true. Martin said in the closing keynote, everything we failed to solve in the 00:25:56.200 --> 00:25:59.848 past decades bites back now. And so many 00:25:59.848 --> 00:26:04.500 challenges that we just did not properly solve from 00:26:04.500 --> 00:26:11.200 access assignment, access review, access governance, but also identification at scale. 00:26:11.200 --> 00:26:17.600 This is really something that we now have to solve quickly and the hard way. Right. 00:26:18.400 --> 00:26:20.965 Yeah, absolutely. I mean, I've mentioned 00:26:20.965 --> 00:26:24.300 earlier that agentic AI behaves like a human, but is 00:26:24.300 --> 00:26:31.734 missing the right or wrong idea or ethics 00:26:31.734 --> 00:26:40.800 behind that. So an agentic AI behaves like a human 00:26:40.800 --> 00:26:44.906 identity inside a threat challenge earlier with 00:26:44.906 --> 00:26:49.100 a proper approach, we would be able to, at least 00:26:49.100 --> 00:26:52.500 to a certain degree, to handle agentic AI much 00:26:52.500 --> 00:26:55.900 better. I mean, we would still have the issues 00:26:55.900 --> 00:27:00.159 with the delegation, but at least we would have 00:27:00.159 --> 00:27:04.600 a concept like zero trust in place. We would have 00:27:04.600 --> 00:27:12.000 the issues of DLP solved to a certain degree, and we would have dynamic authorization 00:27:12.100 --> 00:27:20.200 in real time. Stuff that we could use right now to solve the agentic AI problem. 00:27:20.400 --> 00:27:25.300 But what we can still see is that a lot of organizations are still struggling with 00:27:25.800 --> 00:27:33.700 I would call it the basics of IAM to a certain degree. We are still talking about 00:27:33.700 --> 00:27:41.600 roles and sometimes not even about roles. Organizations are sometimes, not all, 00:27:43.600 --> 00:27:49.800 very immature when it comes at least to authorization. And that is an issue that will 00:27:49.800 --> 00:27:53.869 hit very hard when agents are going rogue. If 00:27:53.869 --> 00:27:58.300 the landscape, the IT environment is not properly 00:27:59.100 --> 00:28:06.100 prepared for an insider threat that behaves like a human, that can strike very hard. 00:28:06.300 --> 00:28:15.300 Yeah, I would fully agree. And the question is really not, is this a separate 00:28:15.300 --> 00:28:18.600 challenge that we need to solve for these NHIs? 00:28:18.600 --> 00:28:21.900 I think the same challenge will come back to us 00:28:21.900 --> 00:28:27.800 also for the identities that we are used to. Privileged accounts, just regular employee 00:28:27.600 --> 00:28:34.400 accounts. When it comes to those who are using our systems from the outside, our partners, 00:28:34.600 --> 00:28:38.721 our service providers that do something within 00:28:38.721 --> 00:28:43.200 our systems that are no longer within our firewall 00:28:43.200 --> 00:28:51.000 delimited area. So we really need to be very fast. And if we would have done that earlier, 00:28:51.300 --> 00:28:56.900 we could have the time now to focus on what's really new. Now we have to clean up the old 00:28:56.900 --> 00:28:59.540 and the new stuff, which really is a 00:28:59.540 --> 00:29:03.500 challenge in itself. The good thing is there will be a 00:29:03.500 --> 00:29:10.800 lot to talk about at EIC 2027 because these topics won't go away. We will look at these 00:29:11.000 --> 00:29:13.550 topics, of course, also, and this is the short 00:29:13.550 --> 00:29:16.100 commercial break at our Identity Fabric Impact 00:29:16.100 --> 00:29:18.383 Days and the Impact Days in general, three 00:29:18.383 --> 00:29:21.100 coming up later this year. We're just starting the 00:29:21.200 --> 00:29:24.079 plannings. If you want to contribute a talk, this 00:29:24.079 --> 00:29:26.900 is not to Philipp, also to Philipp, but not only 00:29:26.900 --> 00:29:31.500 to Philipp. If you want to contribute a talk, please reach out to us. We're just currently 00:29:31.500 --> 00:29:33.518 planning the agendas. So if you have 00:29:33.518 --> 00:29:36.600 interesting things to say about these topics, go to our 00:29:37.400 --> 00:29:40.250 website and go to the call for speakers. We are 00:29:40.250 --> 00:29:43.100 just right now there. So there is a lot more to 00:29:43.100 --> 00:29:46.638 talk about. And end of the commercial break is, 00:29:46.638 --> 00:29:50.100 of course, the sentence. If you want to adjust 00:29:50.100 --> 00:29:57.600 your existing IAM solution towards a more mature, a more capable, a more NHI, a more 00:29:57.600 --> 00:30:01.391 agentic AI capable system, maybe you reach out 00:30:01.391 --> 00:30:05.100 to Philipp and Ormi to receive some guidance, 00:30:05.100 --> 00:30:09.647 some support, some advisory. Final question, 00:30:09.647 --> 00:30:14.400 what was the most interesting talk that you've 00:30:14.400 --> 00:30:20.100 seen? I will answer that as well later. The question is, when you look back at 60, 00:30:20.300 --> 00:30:24.292 if I remember correctly, sessions that we did in 00:30:24.292 --> 00:30:28.200 five tracks over four days at EIC and Alexander 00:30:28.200 --> 00:30:34.000 Platz, what would be the one that you would recommend to everybody to have a look at it? 00:30:34.000 --> 00:30:36.918 To be honest, I'm not sure if I should answer 00:30:36.918 --> 00:30:39.900 that and pin it down to a single presentation. 00:30:41.900 --> 00:30:51.100 But one moment that I remember back was Patrick Parker's presentation and how he 00:30:52.300 --> 00:30:59.800 stated the comparison to the data-centric defense model, I would even call it, to say 00:30:59.800 --> 00:31:02.800 when we want to protect against agents, we don't 00:31:02.800 --> 00:31:05.800 need to focus on the agents, we need to focus on 00:31:05.800 --> 00:31:10.057 the data and ensure that we secure our data. 00:31:10.057 --> 00:31:14.700 He made that very easily understandable with his 00:31:14.700 --> 00:31:18.289 example about the vaults and said when we have 00:31:18.289 --> 00:31:21.800 27 vaults in Germany or in Europe, it is much 00:31:21.800 --> 00:31:25.872 easier to protect 27 vaults than discovering 00:31:25.872 --> 00:31:30.500 100 identities and finding the 30 rogue identities 00:31:31.100 --> 00:31:34.732 in that bunch of identities. That is something 00:31:34.732 --> 00:31:38.600 that I remember very clearly and that was a thing 00:31:38.600 --> 00:31:43.352 that made me think about how to approach agentic 00:31:43.352 --> 00:31:48.400 AI. If we are doing the correct thing with everyone 00:31:48.400 --> 00:31:58.000 saying discovery first or is it not discovery first, is it data-centric defense first? 00:31:58.100 --> 00:32:00.772 Yeah, I think that's an important point and this 00:32:00.772 --> 00:32:03.500 is really, really forward-looking. There are lots 00:32:03.500 --> 00:32:07.035 of great efforts going on also in making the right 00:32:07.035 --> 00:32:10.500 decisions at the right time and then prioritizing 00:32:10.500 --> 00:32:13.144 applying risk-based principles. I think that is 00:32:13.144 --> 00:32:15.900 an important part. If I would have to answer that 00:32:15.900 --> 00:32:18.745 question, I would also not pinpoint one single 00:32:18.745 --> 00:32:21.900 presentation, but there were a set of presentations 00:32:21.900 --> 00:32:29.200 by practitioners who have just shown what improvements they've made. Things that we 00:32:29.200 --> 00:32:32.101 presented three years ago at EIC as the next 00:32:32.101 --> 00:32:35.200 big thing are now implemented, are now working. 00:32:36.000 --> 00:32:44.900 Think of pass keys, think of identities for cargo rather than for people. Implementing 00:32:44.900 --> 00:32:52.800 identity management at scale at a national automotive company. There are lots of great 00:32:52.800 --> 00:32:56.450 talks that are not that super shiny but are so 00:32:56.450 --> 00:33:00.100 to the point when it comes to solving everyday 00:33:00.100 --> 00:33:06.000 issues for large corporations. If I mention a few, BASF had a great presentation about 00:33:06.000 --> 00:33:11.700 how to do access governance. We had BMW who talked about their approach towards identity 00:33:11.700 --> 00:33:17.400 access management and DB Schenker was awarded with a prize when it comes to assigning 00:33:17.400 --> 00:33:20.335 identities to non-humans for a very different 00:33:20.335 --> 00:33:23.400 purpose, but still identities. Very interesting 00:33:23.400 --> 00:33:28.800 talks. I would really recommend that if you have the chance to watch them afterwards, 00:33:29.100 --> 00:33:41.700 please do. There's a lot in there to learn from and we learn from that as well. That's the 00:33:41.700 --> 00:33:48.000 our experience and our research. Philipp, thank you very much for being my guest today. 00:33:48.600 --> 00:33:55.000 This was one very individual, very personal look back at EIC. Others would 00:33:56.300 --> 00:34:03.100 of course highlight very different aspects of that, but that's what this podcast is about. 00:34:03.200 --> 00:34:11.000 We shared what we think is important. Some topics just dictated them into our agenda 00:34:11.000 --> 00:34:14.359 themselves with their own agency, of course, 00:34:14.359 --> 00:34:18.100 and others I think we need to make sure that they 00:34:18.100 --> 00:34:21.550 are not forgotten when it comes to understanding 00:34:21.550 --> 00:34:25.000 the big picture of IAM. Again, thank you Philipp 00:34:25.100 --> 00:34:27.869 and looking forward to having you soon again. 00:34:27.869 --> 00:34:30.700 Thank you Matthias and also thank you for your 00:34:30.600 --> 00:34:34.508 insights. I'm interested to see in 11 months at 00:34:34.508 --> 00:34:38.500 the next EIC what really realized and how people 00:34:38.500 --> 00:34:44.900 solved zero trust in one year and catching up with everything else that we mentioned. 00:34:45.500 --> 00:34:48.896 That will truly be an interesting year until 00:34:48.896 --> 00:34:52.600 the next EIC. If you do that, let us know. Maybe 00:34:52.600 --> 00:34:55.322 there's an award for you next year. Let us know. 00:34:55.322 --> 00:34:58.100 Thanks Philipp. I'm looking forward to having you 00:34:58.300 --> 00:35:01.100 soon. Bye-bye. Thanks. Bye.