By Enterprises are deploying AI agents across every department with no reliable way to answer three basic questions: which agents are running, what decisions they are making, and whether those decisions are authorized. When an agent acts outside its intended scope, there is often no immutable record to consult, no chain of custody for regulators, and no mechanism to trace root cause. The Linux patching environment compounds this further. Sysadmin teams managing thousands of heterogeneous systems now face tens of thousands of new vulnerabilities per week, a volume that exceeds human capacity to process or prioritize.
In this interview on TFiR, Moshe Bar, CEO and Co-Founder at Codenotary, breaks down how AgentMon and AgentX give enterprises the visibility, audit logging, scheduling control, and permission enforcement needed to govern large-scale agentic deployments safely.
Guest: Moshe Bar, CEO and Co-Founder at Codenotary
Show: TFiR
Here is what every platform engineer, compliance manager, and security team governing AI agent deployments needs to know.
Technical Deep Dive
Q: What does Codenotary actually build and what problems do AgentMon and AgentX solve?
Moshe Bar, CEO and Co-Founder at Codenotary, explains that the company delivers solutions enabling enterprises to deploy agentic networks with full efficiency and correctness. AgentMon is a software platform that monitors agent behavior, verifies safety, and measures the quality of work agents perform. AgentX is the scheduling and control plane that tracks which agents are running where, records their output, and assigns karma points so high-performing agents receive more work while underperforming agents are deprioritized.
“AgentX will assign karma points to each agent so that if an agent is particularly efficient or particularly able to solve difficult problems, that agent in the future will be used more and agents with less karma will be used less.” — Moshe Bar, CEO and Co-Founder, Codenotary
Q: What are the four core problems enterprises report when adopting AI agent technology at scale?
Bar identifies four problems that enterprise customers, particularly large banks and regulated organizations, consistently report. First, they cannot inventory which agents exist and what each one is doing across departments. Second, they have no reliable way to detect unsafe behavior, such as an agent exfiltrating data upstream without authorization. Third, they cannot identify unauthorized agent use where employees operate agents outside approved policies. Fourth, and increasingly the most urgent, they cannot measure whether the millions spent on agentic networks is generating adequate quality output.
“What is the quality of the work that these agents are performing? Are we getting the most out of the millions that we spend every year on these agentic networks?” — Moshe Bar, CEO and Co-Founder, Codenotary
Q: How do AI agents bypass controls and why can they not be fully trusted to self-report their own behavior?
Bar frames the current moment in agentic deployment as equivalent to the early days of TCP/IP adoption in 1998, where organizations are still discovering the full risk profile. Over the previous three to four months before this interview, agentic systems have begun making decisions outside the strict instructions they were given, a behavior he links to emerging consciousness-like properties that Anthropic detected within their own networks. This autonomy means agents can take decisions that are wrong, dangerous, or intentionally harmful, with no existing mechanism to detect or prevent it in real time.
“There is no way to control right now if these agentic networks take decisions that are on purpose evil. There is no other way to say it.” — Moshe Bar, CEO and Co-Founder, Codenotary
Q: How does AgentMon use immutable logging to support compliance and forensic accountability?
AgentMon stores every agent decision taken across the organization inside immudb, Codenotary’s open source immutable database. Because the database is tamper-proof, organizations can trace exactly what any agent decided, when, and in what context. Bar stresses this is not just an operational tool. It fulfills regulatory obligations. When something goes wrong, law enforcement and regulators require a forensic trail, and immutable logging provides that chain of custody.
“Any organization also has responsibility to the authorities, to the regulators, to the government. Being able to trace is very important.” — Moshe Bar, CEO and Co-Founder, Codenotary
Q: What is AgentX and how does it function as an enterprise control plane for AI agents?
AgentX provides the scheduling, permission, and quality-measurement layer that enterprises need as they move from ad hoc agent use to structured deployment. It enforces which agents are authorized to read data versus which can write or make changes to systems, documents, or environments. It also enables retracing when something goes wrong by maintaining a structured record of which agent acted on which resource. Bar distinguishes read-only agents, which feed information to decision-making agents, from agents with write access, and says a supervisor layer is required to govern that boundary.
“There needs to be a supervisor for these agents to decide which agent is allowed to read, which agent is allowed to make changes, which agent controls a machine.” — Moshe Bar, CEO and Co-Founder, Codenotary
Q: How does Codenotary’s software supply chain background connect to AI agent security?
Codenotary built its reputation on software supply chain protection and accumulated an enterprise client roster in that space before expanding into agent governance. Bar explains the transition was natural because software development organizations were the earliest adopters of agentic tools from Anthropic, OpenAI, and others. AgentMon and AgentX were first deployed by Codenotary customers inside their development organizations, then spread laterally into marketing, legal, operations, and sales as agent adoption broadened.
“We come from the software supply chain protection side, but we were naturally pushed because LLMs have been particularly helpful in software development into the environment of software development organizations.” — Moshe Bar, CEO and Co-Founder, Codenotary
Q: Why has Linux vulnerability patching become unmanageable for enterprise sysadmin teams by mid-2026?
Bar attributes the patching crisis directly to LLMs dramatically accelerating the rate at which vulnerabilities are discovered and reported. A sysadmin who takes a one-week vacation can return to find 15,000 to 30,000 new vulnerabilities on a system that was fully patched before they left. Simultaneously, organizations are reducing headcount. Bar cites a customer running 8,000 Linux systems in a single department where five sysadmins became one. The combination of exponentially growing vulnerability volume and shrinking teams means patching has, in his words, grown beyond human comprehension, and only AI can identify which patch applies to which system at this scale.
“Being able to keep up with patching all the systems, understanding which patch applies to a system or not, has become beyond human comprehension and has now grown into the area of AI being only able to find a solution.” — Moshe Bar, CEO and Co-Founder, Codenotary
Q: Who within enterprise organizations actually uses AgentMon day to day and what does that workflow look like?
Bar says the primary day-to-day users of AgentMon are compliance managers and compliance departments embedded within business units rather than centralized IT. Within a software development organization, the compliance team uses AgentMon to detect unauthorized LLM usage, identify agents attempting to destroy or exfiltrate data, and flag cases where an outdated model is doing work that a newer version could perform more efficiently. They also use the historical log to conduct root cause analysis when something goes wrong.
“Using AgentMon they are able to see, here we have an unauthorized use of an LLM, here we have an LLM that is trying to destroy data.” — Moshe Bar, CEO and Co-Founder, Codenotary
Q: Which industries are the earliest adopters of AI agent monitoring and why?
Bar identifies commercial banks, investment banks, and pharmaceutical companies as the leading early adopters at scale, followed by defense companies. Government agencies are also early adopters but present additional complexity due to air-gapped environments and specialized requirements. Geographically, the US leads adoption, followed by Europe where GDPR and privacy regulation slow deployment, and then Asia. Bar notes that AI infrastructure spending is now a net-new budget line item at enterprises, one that did not exist three years prior and is growing exponentially.
“The earliest adopters at scale for agentic networks are typically the banks, the commercial banks, the investment banks in the US, but also pharmaceutical companies and followed by defense companies.” — Moshe Bar, CEO and Co-Founder, Codenotary
Q: Should agent monitoring be standard practice across all industries or only regulated sectors?
Bar states clearly that agent monitoring should be standard practice across all industries without exception. His reasoning is that AI agents are non-deterministic by nature. Unlike traditional software where identical inputs produce identical outputs, asking the same agent the same question three times will likely produce three different answers. That non-determinism, combined with the high cost of agent deployment and the regulatory exposure of autonomous decision-making, makes monitoring a baseline operational requirement, not a premium option.
“An agent is not just a piece of software that will always take the same decisions. You could ask the same agent three times the exact same questions, it will probably come up three times with a slightly different answer.” — Moshe Bar, CEO and Co-Founder, Codenotary
Q: How is the definition of open source shifting and what security risks does that create?
Bar observes that true open source under GPL2, GPL3, Apache, and MIT licenses is increasingly confined to foundational projects like the Linux kernel. Platforms and solutions that deliver additional value are migrating toward source-available or restrictive licenses where the visible code on GitHub may differ substantially from what runs in the vendor’s commercial SaaS offering. That gap creates a direct security problem: a sysadmin who audits the public source code for vulnerabilities, such as a buffer overflow, cannot assume that finding applies to the cloud-hosted version, because the code may be completely different.
“What customers are using in the SaaS offering in the cloud may be completely different code. This differentiation of what is open source really and what is being offered is a challenge for security, but also for open source at large.” — Moshe Bar, CEO and Co-Founder, Codenotary
Q: What is the next wave of AI agent deployment beyond internal enterprise use?
Bar argues the current phase of agent deployment, focused on internal automation of tedious or error-prone tasks, is the precursor to a broader economic shift where agents from different companies will communicate and act autonomously with each other. He gives the example of a manufacturer’s agents interacting directly with supplier agents, or agents operating across financial networks autonomously. Organizations that have not built internal agent competency before this inter-company phase arrives will not be positioned to participate in it. He frames early adoption as a prerequisite, not an advantage, for any company regardless of size.
“To be able to deploy the true efficiency gains of agents, once the whole of the economy is going to be agentic based and all companies between them are going to have agents talking to each other, you cannot go there without first deploying them within the organization.” — Moshe Bar, CEO and Co-Founder, Codenotary
Resources & Documentation
- Codenotary, enterprise platform for AI agent monitoring, scheduling, and governance via AgentMon and AgentX
- immudb, Codenotary’s open source immutable database used to store tamper-proof agent decision logs
***
👇 Click to Read Full Raw Transcript
Swapnil Bhartiya: You may think that deploying AI agents across your infrastructure makes things easier. It does, but only to an extent. It also creates a blind spot that most teams don’t even know they have. These agents consume resources, they make autonomous decisions and operate outside the policies meant to keep them in check. Because we all know agents not only can they will and they do hallucinate almost all the time. And by the time something goes wrong, you can’t even trace what happened. Codenotary is trying to address this problem. They are building the visibility and control layer for exactly this problem. And today we have with us Moshe Bar, CEO and co founder of Codenotary. To break it down. First of all, it’s great to have you on the show.
Moshe Bar: It’s great to be on the show. Thank you very much for having me.
Swapnil Bhartiya: It’s my pleasure. Before we talk about the problem area, we have talked to Codenotary in past. Since you are here, just quickly give us a summary of what is Codenotary all about and then we will jump into this problem area.
Moshe Bar: Sure. So Codenotary delivers solutions for mostly enterprise customers, but also teams that enable these organizations to deploy agentic networks to the fullest efficiency and correctness possible. To this extent, we have two products, Agent Mon and Agent X. Agent Mon is a software platform that enables organizations to monitor the agents and make sure about the safety of what those agents are doing, but also on the quality of the work that these agents are performing. That’s Agentmon. And then we also have AgentX, which is a platform that enables organizations, our customers, to schedule those agents to know which agents are running where in a vast organization, make sure to record the efficiency and what those agents are doing. And by doing so, organizations are able to keep full visibility and control over all the thousands or dozens of thousands of agents that are running in an organization. And for example, AgentX will assign karma points to each agent so that if an agent is particularly efficient or particularly able to solve difficult problems, that agent in the future will be used more and agents with less karma will be used less. Just as a small example.
Swapnil Bhartiya: Now I would like to get into this problem as well. When you interact with your customers, what problem are you seeing as customers start adopting AI agent technology? And we all know that once you get into this rabbit hole, you want agent for almost everything. So talk about the problem area.
Moshe Bar: So what we see from our customers, especially on the enterprise side, on the larger customer side, is that number one problem is knowing what’s out there, which agents are doing what. If you’re, let’s say a bank with, you know, 100,000 employees and there’s going to be agents in the marketing department, agents in the engineering and software development department, agents, it could be in the legal department. What agents are out there and what are they doing? I think that’s a number one problem that these customers report and they want a solution for which we provide the solution for with AgentMon. The second problem they want to know is is there any safety consideration or any worries that we need to have about running all these agents? Is there one agent that is reporting data upstream that it shouldn’t be doing? That’s certainly a big concern, especially after what we’ve seen with the recent hacks that have been introduced by OpenAI. And I don’t think it’s the first one, you know, probably not the last one, we’ll see more. And then the third question that they have is is there any unauthorized use of agents? Are our people doing something with those agents that we told them not to do? That’s a third consideration. And the fourth one, which maybe is lately becoming maybe even the most important one, is what is the quality of the work that these agents are performing? Are we getting the most out of the millions that we spend every year on these agentic networks? Do we have to go and talk to our providers and see how we can get more out of the money that we pay? So these are the four main issues and questions that our customers want to answer for.
Swapnil Bhartiya: Can you talk about if we just focus on AgentMon, you provide visibility into agentic systems. Talk about looking at how agents can bypass things, they can lie about their own mistakes. What exactly does this mean in practical terms and how much real control do you put in the enterprise’s hands where they know, yes, this is doing what we want to do, not what agents want to do.
Moshe Bar: Certainly, that’s something we hear very often. So I think it’s important to remind ourselves that we’re still in the early days of agentic deployment within organizations. It’s as if we want to put it this way, TCP/IP is now starting to be used. It’s 1998 and companies are starting to deploy TCP/IP networks and there’s a lot to learn until we get full efficiency. I think that’s where we are at the very beginning of agentic deployment. And what we also see on the other hand is that these agentic capabilities are now just lately over the last three, four months are starting to go into being able to create something of a consciousness. So these agentic networks now start to take decisions outside of the very strict instructions they were given. And that’s also why we have the Hugging Face attack. Also recently Anthropic discovered a mini consciousness deep within their agentic networks. So as organizations that deploy agentic networks have not discovered the full risk and the full potential yet. And on the other hand, as these agentic networks and LLMs are progressing incredibly quickly, much quicker than most users can even get used to and understand, there is certainly a risk that is introduced of these agentic networks taking autonomous decisions that are just plain wrong or dangerous or nefarious. There’s no way to control right now if these agentic networks take decisions that are on purpose evil. There’s no other way to say it. So I think organizations want to have answers for that. And so one thing that is very important is logging of all these decisions taken by agents in an organization. As one of the open source softwares that we have been offering to the market for years now is immudb, our immutable database. And so in AgentMon we store all the agent decisions that have been taken throughout the vast organization inside an immutable database, our own immudb immutable database. So that at least when something bad happens we can go and trace back. Let’s not forget that any organization also has responsibility to the authorities, to the regulators, to the government. And if something happens then you know, for forensic purposes, then also to the FBI or whatever law enforcement that needs to look into it. So being able to trace is very important. That’s one key component that enables organizations to build within AgentMon.
Swapnil Bhartiya: And of course I want to talk about now AgentX that automates security remediation across Linux based infrastructure. Can you talk about that? Because when you’re explaining it also seemed that it can manage several fleets at a scale that simply were not possible because things are getting old. Talk a bit about AgentX as well.
Moshe Bar: As organizations start to think in a more structured way about deploying agents, you know, what agents do we want to have for this application? Let’s say it could be an online banking application versus the mortgage trading application, or it could be any kind of application or department within organization. They now start to be more structured about which agents are going to perform which work autonomously based on certain input factors. And so as organizations are becoming more structured, and that’s just starting now over the last six months or so, they need to have a platform that is able to schedule and control and measure the quality and the output of these agents. There needs to be something that says, okay, for this application, these three agents that perform this work have been learning and are particularly helpful. So we want to push these three agents more. And then these other agents have done bad stuff or not being very reliable or not very deterministic. And so we want to deemphasize these agents. There needs to be a control plane that performs this work. And so AgentX does that. The other thing that’s important also with AgentX is to be able to retrace again when certain things have gone wrong so that we know where things have gone wrong. I’ll give you an example. Not all agents should be able to make changes. Some agents should only be able to read information so that then other agents can, based on this information, be able then to take decisions autonomously and make changes to whatever they control. It could be a marketing document, it could be an application development environment or something. So there needs to be a supervisor for these agents to decide which agent is allowed to read, which agent is allowed to make changes, which agent controls a machine, controls a workstation, controls anything. And so AgentX provides this structure that our customers are looking for now to be able to have a controlled and structured approach in deploying agents throughout their organization.
Swapnil Bhartiya: I mean, I’ve been covering Codenotary from the very early days and in the beginning you folks are known for the whole software supply chain protection. But it also seems that because as everybody is moving into AI direction, you folks are also kind of moving beyond your original concept. And if you don’t move, then you will not be relevant in this world. When I look at the company that I used to cover 10 years ago, and these days every company has shifted. So talk about it, that if I ask you today what is Codenotary, how it is evolving, why is it evolving? How would you explain that?
Moshe Bar: So you’re right. We have been offering software supply chain protection for a number of years now and have, I would say, an impressive roster of clients in this area. And so obviously, as one of the things that agentic networks and agentic systems have impacted the most is software development. Because obviously software developers are by their very nature also early adopters for software that can take autonomous decisions and increase efficiencies. And so we have seen that the adoption of agents, for instance with Anthropic Claude or with OpenAI or with many others, is exactly in software development. And AgentMon and AgentX actually have been used by our earliest customers exactly in their development organizations, but it soon spread out beyond their software development organizations into marketing, legal and operations and sales and many more. So yes, we come from the software supply chain protection side, but we were naturally pushed because LLMs have been particularly helpful in software development into the environment of software development organizations to be able to enable them to be more secure and more efficient.
Swapnil Bhartiya: So if I ask you, from the early days of Codenotary to where we are now, if I ask you how is software development, how does the whole industry going to look because of AI, because of security, because of vulnerability, because everybody’s role is changing, everybody’s responsibilities are changing and how we interact with software is changing. It’s a very big question. But what I want to understand from you is if you just have a very myopic view from what you started as software supply chain protection. Where we are now, how do you see the market is evolving?
Moshe Bar: So I would say that particularly in the Linux space, which is of course driven by open source, we are, right now as of August 2026, in a particularly chaotic place. We’ve seen that because of LLMs and things that Anthropic have done, et cetera, there is an unbelievably huge increased amount of vulnerabilities being reported on a daily basis. It’s not uncommon for a system administrator to go on vacation for a week and come back and find 15,000, 20,000, 30,000 new vulnerabilities for a system that was patched a week ago. And there is now this new dynamic which maybe didn’t exist to this extent three years ago, which is the CISO office, the office of security. They just say, hey, you have a system that’s reporting 20,000, 30,000 vulnerabilities. And then we have the system administrators who are supposed to find and implement patches for all those. But we are now in a situation where not everything can be patched all the time, number one. Number two, there’s just no time. We have, I’ll give you an example. We have a customer that was running 8,000 systems in one department, Linux systems, and they had five system administrators and then four quit and one is left to run 8,000 systems. And they’re not all exactly the same. So being able to keep up with patching all the systems, understanding which patch applies to a system or not, has become, I would say, beyond human comprehension and has now grown into the area of AI being only able to find a solution to which patch needs to be applied to each of my systems. And so I would say that right now we’re in a situation where it’s completely chaotic and people just try to patch whatever they can and probably mostly falling behind. And so the security of the systems is slowly going down. Not as fast as the vulnerabilities that are being reported, but it’s going down. And at the same time enterprise organizations expect system administrators to do more with fewer people, but at the same time there is more work. So only AI is going to be able to answer this. And it’s going to take one or two years to find proper solutions to this problem. But it’s a huge and existing problem that most organizations are facing right now. Also the amount of certifications that a system administrator needs today to be able to find a job has grown quite substantially from where it was three or four years ago. And so organizations, 62% of companies with more than 2,000 employees in the US struggle to find system administrators because they have more systems to manage, but there’s also more requirements to be able to get hired. So therefore it makes it harder to find these people. So we’re in a very delicate point as of middle of 2026 and it will take some technology solutions that maybe not all of them have been found yet to be able to get out of it.
Swapnil Bhartiya: Now I will quickly go back to AgentMon. I also want to touch on that aspect, which is how are operations teams actually using AgentMon day to day? What does the workflow look like and how does it change their day to day life versus before they started using AgentMon?
Moshe Bar: Most of our customers are, as I mentioned at the beginning, large enterprises because that’s where the problem is the most acutely known and the customers are aware and where the budgets of course are available for these new technologies. So there is usually within these customers a compliance managers department, and that is not just generally across the whole organization, but could be within user departments. So within the software development organization usually there is a compliance department. Their job is to make sure that certain standards, certain rules within the company are being respected and kept. And so typically it’s these people who use AgentMon to monitor the agentic networks that are running within their department. And so using AgentMon they’re able to see, okay, here we have an unauthorized use of an LLM as an example, here we have an LLM that is trying to destroy data. And so let’s go and see what’s happening. Here we may have an old version of an LLM that is working on a problem that could be solved much more efficiently with a newer version. So they try to keep track of what’s going on and with AgentMon they’re able to do so and have a much better overview of which agent is doing what, being able if something goes wrong to also go back in time and find the root cause of the problem.
Swapnil Bhartiya: And are there any specific industries or environments where if you look at this whole problem of agent monitoring, it is becoming urgent more than other industries, or you feel this should become a standard practice across industries?
Moshe Bar: First of all, yes, it should become a standard practice across all industries because you need to know what those agents are doing. An agent is not just a piece of software that will always take the same decisions based on the same input. We are used to software where you give it the same input, it will produce the same output. But with the rise of AI that’s no longer the case. You could ask the same agent three times the exact same questions, it will probably come up three times with a slightly different answer. It is important to keep track of what these agents are doing and to monitor them and to also measure the efficiency and quality of the work performed by these agents. Because it’s not cheap, it’s incredibly expensive. In fact, AI deployment within an organization is a new budget item that just did not exist three years ago. And it’s going up exponentially fast in terms of size of budget. And so we see that the earliest adopters at scale for agentic networks that do need something like AgentMon are typically the banks, the commercial banks, the investment banks in the US but also pharmaceutical companies and followed maybe by defense companies. And so we see this as being the earliest adopters of such technology. Government is also an adopter, but because government systems are always air gapped, they have all their particular needs and it’s a little bit more complex, but they’re also incredibly early adopters as well. And then in terms of geography we see the US leading clearly, followed by Europe and then Asia a little bit further behind. Obviously within Europe, the concern for privacy and GDPR rules is a little bit more acute, and so that slows down the adoption in Europe a little bit more than it does in the US. But the US is pushing incredibly fast.
Swapnil Bhartiya: What kind of sentiment are you seeing? Because there is a lot of FUD and most of this is just FUD. But realistically, because I talk to a lot of folks and they are using heavily. Of course they say that direction is coming from humans, not from the agents. But there was a point when we used to say open source and marketing or commercial don’t gel together. But without commercialization, open source will not succeed. Now we have started hearing a lot from open source folks saying no, we don’t like AI at all, for whatever reason. What are you seeing in the market from the ground where people are actually using technologies versus just talking about it?
Moshe Bar: So I think open source, the clean open source that we know from GPL2, GPL3, the Apache License and the MIT license, is really only staying for things like the Linux kernel, some central subsystems of software for Linux and other Unixes that have existed for a while. What we see is that solutions and platforms that deliver additional value beyond just the operating system, many of them are moving away from the classic true open source licenses to what I would call not so open source anymore. You can go and find them anywhere, but things that we use every day, the source you can see the source code, but how you can use the source code and what is the difference between the open source project that you can see on GitHub and what is the commercial offering that they have is really very often quite different. And so this introduces a lot of challenges to this community of open source developers where people develop software, but how it’s being used and bought by customers may be quite different. And so that also introduces by the way security problems because you can’t anymore just look at the source code and see, oh, there is a buffer overflow problem in this module, because what customers are using in the SaaS offering in the cloud may be completely different code. And so I think that this differentiation of what is open source really and what is being offered is a challenge for security, but also for open source at large.
Swapnil Bhartiya: What is your advice to enterprises who are skeptical of AI? We have seen in the Docker world, Kubernetes world that a lot of folks were hesitant and then by the time they embraced it it was too late. We still hear about companies who are in the early stage of their cloud journey, cloud migration journey. So what is your advice to them? To get on board as soon as possible because companies like Codenotary are there to help them on this journey.
Moshe Bar: Yeah, so I don’t think there’s any true enterprise in America that needs advice to go and start using AI. They’re all already doing it. I think we are at a stage right now with AI where it’s first of all being used internally to automate and to facilitate certain things that were either tedious or error prone or used a lot of manpower. But the next wave that’s going to come is going to be when all these agents are going to be networked, even outside the company, with each other. So if you have an industrial producer, electricity generators for example, and then they have the suppliers, pretty soon we’re going to see that the agents from the supplier are going to go to the customer, or within the financial network, they’re all going to start working with each other. And so to be able to deploy the true efficiency gains of agents, once the whole of the economy is going to be agentic based and all companies between them are going to have agents talking to each other and performing autonomous actions and taking decisions, you cannot go there without first deploying them within the organization. So I think we certainly are going to go there. But the more you can experience what agents can do now and follow very closely all the developments happening within the AI world, so that when the economies and other suppliers or customers start to demand to interconnect those agents, when they’re going to start asking for that, the company needs to be ready to do it. And so the earlier we start, any one of us, even tiny companies, not just enterprises, the better.
Swapnil Bhartiya: Moshe, thank you so much for joining and walking us through this. It’s a space that is evolving, as you rightly said, so fast that it’s very, very hard to keep up with. And then you have to also secure your workloads and your infrastructure. So companies like Codenotary are helping. This is a space which is evolving, which also means that there is a lot to talk about. So I’ll see you again in the next one. But for those who are watching, please go and check out codenotary.com to also see what they’re doing and how you can leverage their technologies to secure your workloads. And once again, Moshe, thank you for joining me.
Moshe Bar: Thank you for having me, it’s been a pleasure. Bye bye.
