Croatia: N1 Croatia

20260920 08:00 UTC · 00:30:59 · 504 transcript segments · GDELT Visual Explorer · plain-text transcript · Event Map

Film strip

One frame every 4 seconds, 465 in all. Click a frame to jump the transcript to that moment, or click a line of the transcript to see what was on screen while it was said. Served from apprised.news, so it loads behind proxies that block Google Cloud Storage.

00:00:00
00:00:04
00:00:08
00:00:12
00:00:16
00:00:20
00:00:24
00:00:28
00:00:32
00:00:36
00:00:40
00:00:44
00:00:48
00:00:52
00:00:56
00:01:00
00:01:04
00:01:08
00:01:12
00:01:16
00:01:20
00:01:24
00:01:28
00:01:32
00:01:36
00:01:40
00:01:44
00:01:48
00:01:52
00:01:56
00:02:00
00:02:04
00:02:08
00:02:12
00:02:16
00:02:20
00:02:24
00:02:28
00:02:32
00:02:36
00:02:40
00:02:44
00:02:48
00:02:52
00:02:56
00:03:00
00:03:04
00:03:08
00:03:12
00:03:16
00:03:20
00:03:24
00:03:28
00:03:32
00:03:36
00:03:40
00:03:44
00:03:48
00:03:52
00:03:56
00:04:00
00:04:04
00:04:08
00:04:12
00:04:16
00:04:20
00:04:24
00:04:28
00:04:32
00:04:36
00:04:40
00:04:44
00:04:48
00:04:52
00:04:56
00:05:00
00:05:04
00:05:08
00:05:12
00:05:16
00:05:20
00:05:24
00:05:28
00:05:32
00:05:36
00:05:40
00:05:44
00:05:48
00:05:52
00:05:56
00:06:00
00:06:04
00:06:08
00:06:12
00:06:16
00:06:20
00:06:24
00:06:28
00:06:32
00:06:36
00:06:40
00:06:44
00:06:48
00:06:52
00:06:56
00:07:00
00:07:04
00:07:08
00:07:12
00:07:16
00:07:20
00:07:24
00:07:28
00:07:32
00:07:36
00:07:40
00:07:44
00:07:48
00:07:52
00:07:56
00:08:00
00:08:04
00:08:08
00:08:12
00:08:16
00:08:20
00:08:24
00:08:28
00:08:32
00:08:36
00:08:40
00:08:44
00:08:48
00:08:52
00:08:56
00:09:00
00:09:04
00:09:08
00:09:12
00:09:16
00:09:20
00:09:24
00:09:28
00:09:32
00:09:36
00:09:40
00:09:44
00:09:48
00:09:52
00:09:56
00:10:00
00:10:04
00:10:08
00:10:12
00:10:16
00:10:20
00:10:24
00:10:28
00:10:32
00:10:36
00:10:40
00:10:44
00:10:48
00:10:52
00:10:56
00:11:00
00:11:04
00:11:08
00:11:12
00:11:16
00:11:20
00:11:24
00:11:28
00:11:32
00:11:36
00:11:40
00:11:44
00:11:48
00:11:52
00:11:56
00:12:00
00:12:04
00:12:08
00:12:12
00:12:16
00:12:20
00:12:24
00:12:28
00:12:32
00:12:36
00:12:40
00:12:44
00:12:48
00:12:52
00:12:56
00:13:00
00:13:04
00:13:08
00:13:12
00:13:16
00:13:20
00:13:24
00:13:28
00:13:32
00:13:36
00:13:40
00:13:44
00:13:48
00:13:52
00:13:56
00:14:00
00:14:04
00:14:08
00:14:12
00:14:16
00:14:20
00:14:24
00:14:28
00:14:32
00:14:36
00:14:40
00:14:44
00:14:48
00:14:52
00:14:56
00:15:00
00:15:04
00:15:08
00:15:12
00:15:16
00:15:20
00:15:24
00:15:28
00:15:32
00:15:36
00:15:40
00:15:44
00:15:48
00:15:52
00:15:56
00:16:00
00:16:04
00:16:08
00:16:12
00:16:16
00:16:20
00:16:24
00:16:28
00:16:32
00:16:36
00:16:40
00:16:44
00:16:48
00:16:52
00:16:56
00:17:00
00:17:04
00:17:08
00:17:12
00:17:16
00:17:20
00:17:24
00:17:28
00:17:32
00:17:36
00:17:40
00:17:44
00:17:48
00:17:52
00:17:56
00:18:00
00:18:04
00:18:08
00:18:12
00:18:16
00:18:20
00:18:24
00:18:28
00:18:32
00:18:36
00:18:40
00:18:44
00:18:48
00:18:52
00:18:56
00:19:00
00:19:04
00:19:08
00:19:12
00:19:16
00:19:20
00:19:24
00:19:28
00:19:32
00:19:36
00:19:40
00:19:44
00:19:48
00:19:52
00:19:56
00:20:00
00:20:04
00:20:08
00:20:12
00:20:16
00:20:20
00:20:24
00:20:28
00:20:32
00:20:36
00:20:40
00:20:44
00:20:48
00:20:52
00:20:56
00:21:00
00:21:04
00:21:08
00:21:12
00:21:16
00:21:20
00:21:24
00:21:28
00:21:32
00:21:36
00:21:40
00:21:44
00:21:48
00:21:52
00:21:56
00:22:00
00:22:04
00:22:08
00:22:12
00:22:16
00:22:20
00:22:24
00:22:28
00:22:32
00:22:36
00:22:40
00:22:44
00:22:48
00:22:52
00:22:56
00:23:00
00:23:04
00:23:08
00:23:12
00:23:16
00:23:20
00:23:24
00:23:28
00:23:32
00:23:36
00:23:40
00:23:44
00:23:48
00:23:52
00:23:56
00:24:00
00:24:04
00:24:08
00:24:12
00:24:16
00:24:20
00:24:24
00:24:28
00:24:32
00:24:36
00:24:40
00:24:44
00:24:48
00:24:52
00:24:56
00:25:00
00:25:04
00:25:08
00:25:12
00:25:16
00:25:20
00:25:24
00:25:28
00:25:32
00:25:36
00:25:40
00:25:44
00:25:48
00:25:52
00:25:56
00:26:00
00:26:04
00:26:08
00:26:12
00:26:16
00:26:20
00:26:24
00:26:28
00:26:32
00:26:36
00:26:40
00:26:44
00:26:48
00:26:52
00:26:56
00:27:00
00:27:04
00:27:08
00:27:12
00:27:16
00:27:20
00:27:24
00:27:28
00:27:32
00:27:36
00:27:40
00:27:44
00:27:48
00:27:52
00:27:56
00:28:00
00:28:04
00:28:08
00:28:12
00:28:16
00:28:20
00:28:24
00:28:28
00:28:32
00:28:36
00:28:40
00:28:44
00:28:48
00:28:52
00:28:56
00:29:00
00:29:04
00:29:08
00:29:12
00:29:16
00:29:20
00:29:24
00:29:28
00:29:32
00:29:36
00:29:40
00:29:44
00:29:48
00:29:52
00:29:56
00:30:00
00:30:04
00:30:08
00:30:12
00:30:16
00:30:20
00:30:24
00:30:28
00:30:32
00:30:36
00:30:40
00:30:44
00:30:48
00:30:52
00:30:56

Transcript

Google Cloud Speech-to-Text API (Chirp) + Gemini 2.5 Flash Non-Thinking. Treat it as a searchable index of what was broadcast, not a quotation record.

00:00:00The most important origin, because
00:00:03with us, the main roles are only given to originals from Croatian
00:00:07fields. Domestic in the main role
00:00:10Podravka
00:00:13broken.
00:00:35sebe u firmi, onda sad upravljaju u
00:00:41computers, to printers, to
00:01:05Good day and welcome to another AI central point broadcast.
00:01:08We continue with a new series of episodes in the
00:01:11second half of the year. Today's topic will be AI
00:01:14agents, so when software stops waiting for
00:01:17commands. With me in the studio are my guests today,
00:01:21Nenad Raca, CIO of Aduro Idea, and Krunoslav
00:01:24Kedmenec, program manager at
00:01:32Until yesterday, or even today, we asked artificial intelligence something and it gave
00:01:35answers. Mm, this year we are starting to give it
00:01:39tasks. Are you ready to let software
00:01:42actually do something on your behalf, maybe even without asking you
00:01:46beforehand? Yes, we are, it's already
00:01:49safe enough, ready enough. Okay, great, good, we'll get back
00:01:53to that. Yes, we are doing that currently. You are already doing it, that will be very
00:01:57interesting, I think, for many listeners to hear. They are thinking about it.
00:02:00So, we already have people in Croatia who, without problems, with full confidence,
00:02:04mm, use agents. I actually wrote some
00:02:08articles about agents two years ago and I predicted that 2004,
00:02:1124 and 25 would be the year of agents. I miscalculated a bit,
00:02:15it seems it's only 2026.
00:02:18And today everyone talks about agents, uh, but what are
00:02:21agents actually, what is it actually and how would Nenad describe the difference between
00:02:25a chatbot, automation and actually a true AI agent? In fact, that's
00:02:29the most... important difference. A chatbot answers questions,
00:02:32automation always does the same routine work and there are no
00:02:36changes, and an agent, in
00:02:38quotation marks, thinks, makes decisions, knows when it can do
00:02:42something and when that task should be handed over to a human. So, it has a certain
00:02:46level of responsibility and decision-making in all of this. Hmm, is
00:02:49an agent actually a new technology or is it
00:02:53a new level of trust we have in old, already developed technology?
00:02:56How much has that change happened? That is actually the most important part, that new level of trust. The
00:03:00technology is not that new, all those parts that we are incorporating
00:03:04have existed for several years, but now we have shaped them so that we can
00:03:07really trust them, or rather that mechanisms of
00:03:10trust and verification of all those agents are established. In
00:03:13fact, with agents, we are opening a new dimension where
00:03:17these things actually become our new digital employees. Hmm, hmm. Well,
00:03:20yes, an agent should be seen as another interface between
00:03:23human and computer. Hmm, so
00:03:27you no longer have a keyboard, or if we go to a higher level... you no longer
00:03:30have, you write a specification, then someone who
00:03:33translates it into computer language, no, you now directly communicate with
00:03:37the computer and directly give it a task. So, that's your
00:03:40intern. How did that process actually go at Verso Altima
00:03:44and did that moment happen where you said, okay, this is no longer just
00:03:48a tool, this is already an agent with a certain level? Did that
00:03:52happen, was it a process that took time? Well, in principle,
00:03:55we started last year, each of us individually, using
00:03:58AI for something. What it does for some documents, for
00:04:02some, uh, review of some contracts and so on, but those
00:04:06were islands, it was exclusively communication
00:04:09uh with this LLM,
00:04:13then the analysis of what needs to be uh for us
00:04:16Aduro, when we went to try if we could
00:04:20use AI for
00:04:23cyber, they were the first to come with an agent and it was like wow,
00:04:26it works, that was the wow moment. That was even before
00:04:30Anthropic released it commercially and the moment Anthropic
00:04:33released the personal agent, that was
00:04:37it. So, there is that wow moment when you saw something and despite the experience
00:04:40you said, okay, this is much better than what we expected.
00:04:44- I think it's good to mention Kruno, you actually have an example where agents
00:04:48within the company have already done the work from beginning to
00:04:52end, meaning redesigning some corporate websites. But can you
00:04:55walk us through what the agent did, what the human did? That's
00:04:58a great example where almost anyone can
00:05:01do it. Mm, agents really like
00:05:04uh examples. Hmm, this is excellent. You take
00:05:08an agent and say, these are our corporate websites, this is how I would like them to
00:05:11look. Great,
00:05:14then the interaction begins, which is, let's say, more or less something like chat
00:05:17ideas are elaborated and the agent makes
00:05:20a plan on how we will implement it. Okay, that is given
00:05:24to another agent, who then codes it. Then it goes to
00:05:27transferring that code to WordPress. And that's where it got very interesting, because
00:05:31my colleague gave the agent the task of transferring it to WordPress, but it wasn't quite right. Then he
00:05:34kept reporting, no, this is not good, change this, then he was
00:05:38frustrated and gave the agent the task of
00:05:40comparing what was done and what was on WordPress, and the agent
00:05:44reported back to the other agent what needed to be done, and today we have beautiful
00:05:48websites. So, here's a question that will interest everyone in
00:05:51business. This might also be a good time to mention that we are organizing
00:05:54an artificial intelligence conference in Rijeka for entrepreneurs. As an entrepreneur, I would
00:05:58like to know how long that process took with agents, and how long it
00:06:01would have taken without agents, if only people had done it. What is the time saving?
00:06:06Well, even more than 10 times, I would say that
00:06:09uh days and weeks are... hours.
00:06:13Aha, so what you used to need several weeks to do,
00:06:16have meetings, talk to people who would
00:06:19create graphics for you, you would make
00:06:21icons, check all the texts, that literally took
00:06:25a few days, maybe. Maybe the biggest benefit is not just time, but
00:06:29the fact that we suddenly have the capacity to, instead of working, talk a lot
00:06:33and decide what to do, we make those five versions, look at which one is
00:06:36the best, and practically immediately choose from choices that were made very quickly
00:06:40and efficiently, which are complete, which are not conceptually, and yes, they are not prototypes but
00:06:43practically functional, in this case, websites. So, I would
00:06:47say I think many members of UGP will be more interested in hearing
00:06:51about this. Nenad, let's go back to Aduro Idea. So, you have experience and have been
00:06:55working with agents for some time and actually with agent
00:06:58operational centers. Uh, what is that exactly, how to
00:07:01explain it to someone who is not from IT for starters, and again, let's go back to
00:07:05some of your examples, 10 times is a huge saving, so practically 90%
00:07:09of time is saved. Does that match your experiences? Well, my experience... actually a little
00:07:12less, but because of this playing part, that everyone suddenly has the opportunity to play,
00:07:16or rather, I don't know, sometimes when we were developing software we always talked a lot at the beginning
00:07:20before we started working. And now we allow
00:07:23ourselves to start working on different versions and see which one
00:07:26will be the best. So we have a little more wasted time, but the improvement is still
00:07:30significant in that whole story. What you
00:07:33mentioned about what an agent system is,
00:07:37it's probably best if we consider agents as digital employees.
00:07:40Hmm, the entire operating system for...
00:07:42is actually their building where they work, with their rules,
00:07:46like every company has its own additional company for
00:07:49digital people, for digital agents who work for it, in
00:07:52which physical people can enter, can monitor it, can see how it
00:07:56functions, how it functions, create new agents within it.
00:08:00Hmm, that's perhaps a new thing now. Until now, we all looked at who has which
00:08:03employees, how many employees, how they are organized. Now we will
00:08:07have a new sector that will look at how many employees there are,
00:08:11what they are like, how they are organized, how they work, how efficient they are, how
00:08:14accurate they are. Yes, I mean, that part sounds fantastic, although it still seems a little
00:08:18abstract to a very large number of people in Croatia, so we will definitely get back
00:08:22to that, but here's a question for you, and perhaps for Kruno.
00:08:25So, where in an average Croatian company is an agent
00:08:29used, or where can it be used to be actually
00:08:32most effective, and where are these not some spectacular applications? Well,
00:08:35I mean, there are many of these trivial ones that everyone pushes first, like customer
00:08:39service. Those are things that are very easy for everyone to understand.
00:08:42Easy to understand how it works. But maybe there's that part that all
00:08:46people forget, that they use very often, and that's the part of being a kind of consultant in
00:08:50certain decisions, where we start inserting some
00:08:53documents, communicating, and actually creating
00:08:56a new consultant with whom we can discuss some
00:08:59topics where sometimes people in the company don't have the right interlocutors, so they used to
00:09:03look for them outside. Now, with an agent, they can have an interlocutor with whom they can do that,
00:09:07analyze documents, analyze decisions, try out some
00:09:10scenarios or any such things. A concrete example, you have a large
00:09:14project with many subcontractors, main
00:09:18contracts, contracts with subcontractors.
00:09:21Give the agent the task to read all those contracts and to
00:09:24say, these you have covered, these you have not
00:09:27transferred. So, no, that's not, that's still not
00:09:30a lawyer, but it immediately provides insight, especially agents are
00:09:34very good when it comes to putting together information from
00:09:36different sources and doing it very well. We had an episode about artificial intelligence
00:09:40in law some time ago, and people who
00:09:44work on specialized systems for artificial intelligence in law testified to that. So, we can always
00:09:48refer to that episode. Some
00:09:51people complain again, so I have to ask you, are we at a stage where agents save
00:09:55us as much time as you said, or do they create new work for us in the sense that
00:09:59someone still has to supervise these agents? Uh,
00:10:02I don't think that supervision is a problem, because it's basically
00:10:06like, uh, how would I say, when we had computers, when they arrived, we all
00:10:10became secretaries, and we all
00:10:13started doing that administrative work, and in fact, the jobs of secretaries didn't disappear, but
00:10:17we all became our own secretaries. Hmm, what we are
00:10:20becoming now, we are all actually becoming managers, because we all now get our own
00:10:24digital teams that we start managing. So,
00:10:27when you look at it, it's not taking away jobs, but it's actually defining
00:10:31some new roles for people who perhaps were not ready for them until now. Yes, no, the question is
00:10:35whether new work is being created, because you now have to control agents that
00:10:39you didn't have to control before, but my experience is that the scope of work
00:10:43has not increased, that we are on the winning side overall. The type of work has
00:10:46changed, because we are doing more work, rather than controlling someone
00:10:50or some people who are doing work for us. And people who have
00:10:54managed infrastructure until now, first they managed infrastructure
00:10:56within their company, then now they manage it
00:11:00in the cloud. Those same people will manage the agent infrastructure, they will
00:11:03make sure everything works, that not too many tokens are used, that it
00:11:07functions. But we can imagine, building on what Nenad
00:11:11said, an AI agent as a board advisor, for example, meaning
00:11:15someone who will help in decision-making. I think that's a very interesting part and I agree
00:11:18it's often neglected but potentially very good. But
00:11:21okay, mm, I'd like to ask Kruno one more thing. "Actually,
00:11:25you have a very interesting project, namely, AI for Cyber, which is
00:11:28a research and development project for specific cybersecurity,
00:11:31a topic very dear to me, which is why we had an episode
00:11:35specifically dedicated to it. So, it's being carried out with funding
00:11:38from the European Union, you managed to get that. What problem did you want to solve with agents
00:11:42with this? The problem is, one of the things Verso Altima
00:11:45does is cybersecurity, mhm, not
00:11:49only for IT systems but also for smart
00:11:53city management systems."
00:11:55and industrial control systems and so on. And there you
00:11:58have today's software, today's methodology is
00:12:02pattern recognition, hmm. You constantly
00:12:04maintain a
00:12:07database of event patterns, it
00:12:11recognizes them and gives them to an expert for assessment
00:12:14whether to react. Unfortunately, there are a lot
00:12:18of false alarms, a lot of things
00:12:20that recognize a pattern, but we know it's just some
00:12:24test. Hmm, what we, as I said, what was that
00:12:28wow moment, we gave a hastily put together
00:12:31agent to just go through a bunch of logs and it started
00:12:35spitting out things we hadn't even thought about.
00:12:39So, we are moving from pattern recognition to
00:12:41anomaly recognition, hmm. Is something
00:12:45happening that usually doesn't happen? Am I logging in
00:12:48at 2 AM and mistyping my password 10 times? Hmm,
00:12:52hmm. I have to ask right away, it reminds me of some examples from other
00:12:56industries. How much time was actually wasted until now on
00:12:59sifting through those false alarms,
00:13:03and actually how much has an agent accelerated that and
00:13:06saved time? Here, the approach changes, right?
00:13:10The previous approach was to search for something, but here
00:13:13the agent searches for anomalies in everything that behaves, and
00:13:17the most important thing is, perhaps when people looked at it, the problem was always that we were looking at
00:13:21certain different sources. Here, with agents, it's already possible to look at
00:13:25connections between totally different sources that occur at a certain time and that might have
00:13:29some causal links. So, when something is hidden or
00:13:32disguised within some other events, I assume it's more
00:13:35efficient. Yes, for example, another big project we are working on is
00:13:39the use of agents for monitoring huge telecom networks.
00:13:43Aha, you have very trivial situations where somewhere in
00:13:47Split there is a device that has an old version of
00:13:51Finware. And how do you find that? So, you are looking for what Ned said, you are looking in
00:13:55what no, you let the agent, tell me what's wrong,
00:13:58and he, he nicely says in Split there is a
00:14:01device that has an old one and do you want me to
00:14:04upgrade it? Yes, that's what you'll let
00:14:08the agent do. Today he asks, today he asks you. I think that
00:14:12example will interest many companies, like where we all have to upgrade companies, no,
00:14:15absolutely. Super, Nenad, but what
00:14:18happens, let's go back to that topic. If attackers use
00:14:22agents for some cyber security mm... problems, does
00:14:25the speed of attacks actually change, how important is human
00:14:29defense there, does a human even have time to react? It's not even a question of how
00:14:33fast it is, but how inventive it becomes, hmm.
00:14:36Right, that's a new level, because until now that
00:14:39part of invention was done by special people who had the gift of finding such
00:14:43things, and now with agents, or rather
00:14:47swarms of agents, that means now we are no longer talking about one agent, we are talking
00:14:50about them, which can be hundreds, acting simultaneously, communicating
00:14:54with each other. They can carry out very simultaneous and diverse
00:14:57attacks and diverse testing of different things. So,
00:15:01uh, we are currently in the company using and testing
00:15:04uncensored models to test the security of our
00:15:07applications. So, instead of doing the traditional testing of
00:15:11applications, whether the code is secure enough and everything, we simply let an agent
00:15:15who has permission to attack the application, to try
00:15:18what it can do with the application, because simply new patterns will emerge and
00:15:22new patterns are emerging, new things that work. Yes.
00:15:25Innovativeness is always a specially interesting question. It's interesting in this
00:15:28context to say that just these days, information came out from OpenAI that
00:15:31with the newest model, they managed to solve one of the extremely
00:15:35difficult mathematical problems. In that way,
00:15:38again, suggesting a completely new kind of innovativeness and knowledge, so
00:15:42let's hope it won't reflect in all these other security
00:15:46problems, I think it will, because
00:15:49a colleague, one of my physicist colleagues who uses
00:15:53AI in that whole story, he said just recently that
00:15:56he was getting such solutions, meaning he uses
00:16:00AI very intensively in his work, meaning that he gets solutions where he
00:16:03sometimes has to think hard about the approach that AI took.
00:16:07Which, in the end, very often turns out to be very good, but at first it wasn't
00:16:11visible to him at all. That's the thing, how AI helps us
00:16:14find inventive approaches to something, anomalies. In fact, let's
00:16:18stay on that risk for a second, meaning today, still a small
00:16:21number of companies use agents, but I'm sure that will increase, especially after they hear all of you.
00:16:25An increasing number. But then we ask ourselves, what is that risk, because if we
00:16:29use some agents, we have given them access to our data, to some decision-making,
00:16:33how risky does that actually become for us, is that something companies should
00:16:37be afraid of? I mean, they should be afraid of it just like
00:16:40anything else. Uncontrolled, unorganized access brings
00:16:44problems. If it's approached smartly,
00:16:47organized, so that what is being done and what is being managed is known, what
00:16:50rights are given, because it's not a problem if the agent has the right to what
00:16:54it needs to do, but if it doesn't have the right to what it needs
00:16:58to do, it won't do the job. If it has too many rights, it might do
00:17:02too much. So, it's the same as people, you need to look at its
00:17:05experience, gain experience in... managing, then accordingly give
00:17:09rights, manage rights. Fire is a good servant, a bad
00:17:12master. The same applies to agents.
00:17:16If, as Nenad says, you know very well what you are doing, how you are doing it,
00:17:19what rights you are assigning, there are no problems. If you just let it go,
00:17:23you've thrown a match on dry grass and everything's gone. Okay, but who's
00:17:27to blame, for example, if an agent deletes the wrong email or
00:17:30sends the wrong offer? Always the organization and those who have
00:17:34defined the processes. The software cannot be blamed. Hmm.
00:17:37It does what it's allowed to do. So, no agent will ever do
00:17:41anything outside of what it's allowed to do and what's defined as its
00:17:44task, but it will try with all its capabilities to complete that
00:17:48task. These incidents that are talked about a lot, it was given the task
00:17:51to do that and they watched how it would do it. It surprised them how
00:17:55it did it. Yes, that's this speed, that is, how quickly
00:17:58it can happen. That's this inventiveness and speed with which it can be
00:18:02created. Yes, I think that's a special question that we probably
00:18:05won't go into today, but for example, when we have... many agents on both sides, and
00:18:09so maybe it's not a bad context for my next question. So, we've been researching
00:18:13"human in the loop", hopefully our article on that topic will be out soon. When is
00:18:17a human in the loop important, when is it necessary, and when is it actually
00:18:20unnecessary or impossible for them to do? Because as humans who
00:18:23approve something, we can look at 10 things or maybe 100 things or a few hundred, but
00:18:27if it's 10,000, 100,000, a million cases that we have to,
00:18:31the human brain simply cannot process that. Mm, what are
00:18:34your thoughts on this? Well, we need a human in the loop. That is, a human
00:18:38in the loop, meaning where the human is, I mean, simply so that we don't have them just for formality, that's
00:18:42important, not for formality, but where they are truly needed. So, truly in
00:18:45tasks where the crucial thing is that something happens,
00:18:49that something is done, but there must always be some
00:18:51mechanism for it to be done, because there are many
00:18:55cases where a human will be put in the loop just
00:18:58to perhaps avoid some legal responsibility, and then he'll be blamed,
00:19:02he didn't look, but if he clicks 10 times he will click 11 and
00:19:05111 and 1001.
00:19:08Automated, it will become automated, we have actually made a robot out of a human. Hmm, what
00:19:12I say, just last week I was talking to
00:19:14colleagues, I realized where the bottleneck is
00:19:17in using an e-agent. I, yes, I can no
00:19:21longer keep up with how quickly it can analyze,
00:19:24suggest, ask a multitude of questions from
00:19:28different aspects, I physically can no longer
00:19:31keep up with it, so not just approval, but also the work with
00:19:35the agent itself is too fast for us. I've heard that from... several sides, exactly,
00:19:38exactly that experience. Then the question arises, so on the one hand
00:19:42we have the AI Act which requires human supervision, and
00:19:45then we ask ourselves when that is, is it feasible in practice or is it just some words on paper where they told
00:19:49me that a human is responsible on paper but actually
00:19:53doesn't have the ability to do it because it's a bottleneck? Yes, but again, for that, we need to know how it works,
00:19:57meaning that's again, we're back to the standard thing, enough testing, where again
00:20:01e-agents can help, because we can also use agents to test agents.
00:20:04So, what we said, if they can carry out some
00:20:07attacks, they can do anything else. An AI agent can test an AI agent and
00:20:11find ways, or rather, problems in behavior at points
00:20:14where human supervision is truly
00:20:18justified. Yes, and you teach him while you work with him. So, all this that
00:20:21I'm saying, I can't keep up with it, but every iteration, he understands better and better.
00:20:25In the end, I no longer need him, that's that
00:20:29moment where I let him do what I assigned
00:20:32him, but by then we have solved it. My experience is that it takes
00:20:36about a week.
00:20:39Hmm, yes, I think that's the key part, that agents actually learn from us, and my
00:20:42experience is, lately I've been playing with an AI second brain, so that's
00:20:46a system that is being developed, which has several agents connected, and it's constantly learning from me.
00:20:50It collects data, so I assume that will also be a part that I
00:20:53plan to, let's say, deal with educationally, meaning
00:20:57helping people create all their second brains that will help them, that will learn from
00:21:00them and actually help them be 10 times more efficient or even 100 times
00:21:04more efficient. But here we need to take into account, what you mentioned, these are personal
00:21:08agents, right? A second brain and such things serve me personally. If we go to
00:21:12a corporate level for company use, then that
00:21:15approach no longer applies, because then uh, we can't let someone, in quotes, for... memory, that memory must be
00:21:19controlled. So, access to memory
00:21:23in the corporate world and in the private world is totally different, and that's the
00:21:26thing that I hope people won't make the mistake of trying to
00:21:30use private agents. And what works well privately in corporate
00:21:34is good, so separating my second brain as an agent, as my private
00:21:37multiplier of my efficiency, which is a fantastic thing, and
00:21:41corporate agents who are made and controlled differently and so on, great, that's
00:21:45an important thing for. Yes, that's my experience, I have to
00:21:48separate private matters. Hmm, so if I'm looking for LED lighting
00:21:52for an aquarium, that shouldn't be within the AI for
00:21:55Cyber project. Those are two different contexts, because at the beginning, before I
00:21:59mastered those skills, then I was getting strange answers, and then I realized,
00:22:02well, yes, I asked it some nonsense within the same context,
00:22:06now it concludes that I need an aquarium for cyber. Great,
00:22:10great, let's then give one more, maybe, practical advice
00:22:13to the viewers. A company comes and says, "I want an agent," we're interested,
00:22:17that sounds great. What should they have first, how? To start that process,
00:22:21maybe it has nothing to do with artificial intelligence? Well, first they need to have
00:22:24a clear idea of what they want to achieve. Okay, second, they need to make
00:22:27a decision that they want to change how they want to work. Because
00:22:30implementing AI in a company without changing the way
00:22:34the company works is doomed to failure. Hmm, so those are all
00:22:37projects that fail, they fail because the basic work process has not
00:22:41changed. The basic work process must change because AI is not an addition to
00:22:45the process, it is an evolution of the process itself, which starts working immediately after that.
00:22:49Then we can go... to technology and everything else. That's perhaps
00:22:52the most important message. So, my background is it, al'
00:22:55if it doesn't fall into it, it's clear, have you
00:22:58managed to change the processes and mindset within the film, my advice is to get
00:23:02a personal AI agent who will guide you through this process, I know
00:23:05some people who have done it and are very satisfied with it, I must
00:23:09say even at the level of personal development, what insights they have gained
00:23:12into their functioning and behavior, how much does it cost to get into
00:23:16such projects at all, how much does it cost to get in if you do it smartly?
00:23:20How much does it cost if you do it the wrong way? Let's start with this, if you do it the
00:23:23wrong way, there are those penalties for GDPR European Ai act,
00:23:27that's how much it costs to do it the wrong way, so that's the ultimate
00:23:31price that is paid for all that plus lost time plus lost
00:23:34reputation plus everything else to get in smartly
00:23:38we can already do it for a few thousand euros, mhm, so these are not
00:23:41some abnormally expensive projects, abnormally expensive projects, so the important thing there
00:23:45is that these are no longer projects, so it's not something that is created from
00:23:48scratch, there are ready-made things that exist, that work, that can be
00:23:52tried out, because uh, I don't know, in our approach we always
00:23:55give people access so you can try it out, you can
00:23:59get a demo, you can see how it works before you decide
00:24:02to implement it, we don't have to enter risky projects, let's enter
00:24:05something small, a small process, but we provide support
00:24:09on how to redesign such things, not just let's
00:24:12force it into the process and everyone who does something will have their own
00:24:16AI assistant to ask something, because realistically, then they haven't sped up the work. "I hope
00:24:20you smartly entered that project, uh, of course, first
00:24:24start with something you know well, mhm, mhm, so
00:24:28you can assess whether what the agent is writing to you
00:24:31is good or not, yes, I often play dumb when
00:24:35talking to him, I deliberately won't tell him, I want to test how he understands
00:24:39and I want to see what all I have to tell him, explain to him, what
00:24:42documentation I have to give him so that he can give me a quality answer so that
00:24:46he can assign work, so start with something I know."
00:24:50That's why I said websites, nothing
00:24:52specific, issuing invoices, checking if
00:24:56there are any unpaid invoices in the entire database of excel
00:25:00documents, financial analyses, some reports automatically, so there are a whole
00:25:04bunch of great things, we will continue to explore this, but for example, you
00:25:07are definitely companies you work with and clients are some early
00:25:11adopters, no, we always have those who need a little more who are skeptical
00:25:14about what is happening, what is developing, what would you say to a director
00:25:18who says: "Let's wait another year for
00:25:22it to settle down, and then let's see and maybe introduce an agent, uh,
00:25:25the biggest, biggest problem you can create in business
00:25:28is not following what is happening, that is, waiting
00:25:32too long mhm and not trying things out, so no one says
00:25:36that companies absolutely have to implement this in their business at this moment, mhm, but
00:25:40they have to start using it, but not in a way where we ask chat and
00:25:44chat answers something, but let's try to test some
00:25:47processes, see, we have the opportunity to... learn, technology is
00:25:51new, there is still no one who can say with this
00:25:54technology this, this, this can certainly be done, and this certainly cannot
00:25:57be done, we are all experimenting here, so it must be tested, we are all testing, we are all trying, so
00:26:01all the time delaying. is a waste of time for
00:26:04learning, those who start in half a year, have already lost half a year
00:26:08of learning, those who start in a year have lost a year of learning,
00:26:11because changing the company, what I mentioned, to change the way
00:26:15the company functions is a prerequisite for
00:26:17implementing AI, that is, AI agents, that is, digital
00:26:21employees, opening new departments, new things in the company itself, how it will
00:26:25work digitally, so these are no longer digital software, these are
00:26:28digital employees, so he provides complete work and results and
00:26:32control and supervision over him, super similar processes as
00:26:35it was, digitalization of offices, when we
00:26:39switched from paper to
00:26:43networks, so to make a transformation that, you know,
00:26:47you have to see what the possibilities are, that is, learn what
00:26:50you can do with agents then make decisions and say great, an agent would greatly
00:26:54help us here and start there, oh great, I can
00:26:58only say how lucky we are who deal with education in artificial intelligence,
00:27:02mm... guys, thank you very much, we will continue this conversation in AI Second Point which will
00:27:05be broadcast on YouTube and with this I will
00:27:08conclude this show about agents, they are an extremely interesting
00:27:12topic, we will certainly return to it, and we plan to cover
00:27:16other interesting topics such as artificial intelligence in marketing or artificial
00:27:19intelligence in the defense industry, artificial
00:27:22intelligence perhaps even in religion or journalism, but
00:27:25you will learn all about that in due time. Thank you all again for your
00:27:29attention until the next AI Central Point show.
00:28:00Do you think it's easy to get into Podravka
00:28:03products?
00:28:07You have to be much better than the competition.
00:28:11Congratulations, you got the role, but we admit,
00:28:14the most important thing is origin, because with
00:28:17us, leading roles are only given to originals from
00:28:21Croatian fields, local in the leading
00:28:24role, Podravka. Do you feel
00:28:27insecure on your feet, is it difficult to get up without support
00:28:30if others have to help you? You get up and walk, meaning
00:28:34you depend on others. Ordinary canes are not reliable, they can easily
00:28:37slip out of your hand or slip. We present
00:28:41the smart cane, the most equipped walking cane with two handles that
00:28:45gives you maximum support and stability so you can easily
00:28:48get up. The smart cane has a three times more secure three-legged
00:28:52base that firmly adheres to all surfaces, even wet, uneven
00:28:56and slippery floors. The joint system on the wide base rotates
00:29:00360° and non-slip contact points with the ground.
00:29:03It is so stable that it stands by itself, unlike ordinary
00:29:07canes that fall to the ground immediately if you let go of them, which is
00:29:11dangerous, because then you can also fall and injure yourself. On
00:29:14the smart cane, you can adjust the height, it has five levels of
00:29:18height that extend up to 96 cm.
00:29:21Extend it as much as needed and adjust it to your height.
00:29:25It has a strap that you can put around your arm to feel even
00:29:28more secure. The smart cane can be folded so it doesn't take up space.
00:29:32Always carry it with you, the smart cane has a built-in LED
00:29:36flashlight that you can move, point the light
00:29:39straight ahead to see in front of you, you can also tilt it to
00:29:42illuminate the path for walking in the dark, it even has a button
00:29:46with a security alarm, if you think you are in danger,
00:29:49press it, if you feel unwell, press the alarm and call for
00:29:53help, order your smart cane today at a
00:29:57reduced price, only the smart cane has two handles for
00:30:00double safe support. Height adjustment to five
00:30:04levels, swivel three-legged base for all terrains,
00:30:07stands independently when you need free hands, built-in LED
00:30:11flashlight for safe walking in the dark, security alarm
00:30:14against thieves or if you feel unwell and it also folds so you can easily
00:30:18carry it with you, be independent and order today
00:30:22the most equipped and smart cane at the most affordable price on
00:30:25the market.
00:30:46The triumph of the AFD: how Germany will change, the Black
00:30:50Sea, the Russian-Ukrainian war against food,
00:30:54Nepali guest workers, what Sherpas do in
00:30:57Norway.
Data courtesy of The GDELT Project (gdeltproject.org), from the Internet Archive TV News Archive. Film strip and transcript are GDELT's, rehosted here under their terms of use, which permit it with this citation.