[00:00:00] Chapter 1: Introductions and “Code RED” Premise
Mirko Novakovic: Hello, everybody. My name is Mirko Novakovic. I am co-founder and CEO of Dash0. And welcome to Code RED code because we are talking about code and red stands for request errors and duration. The core metrics of observability. On this podcast, you will hear from leaders around our industry about what they are building, what's next in observability, and what you can do today to avoid your next outage today. My guest is Chen Harel. Chen is a veteran observability expert and consultant, and the former VP of products at Coralogix and also co-founded OverOps, which was later acquired by harness. Happy to have you on board today. Chen.
Chen Harel: Super excited to finally record this episode.
Mirko Novakovic: Yeah, I think we both had a battery issue with our car, right?
Chen Harel: For sure. Even though Mike wasn't electrical, so. Oh, it is a petrol. But, you know, the electricity just died on me. It was very, very brutal.
Mirko Novakovic: Also not good. But that's a good intro to my first question, a Code RED moment. So what was one of your biggest Code RED moments in your career?
[00:01:10] Chapter 2: The AutoCAD Outage and Birth of an Idea
Chen Harel: I wish that was my biggest Code RED moment. Nope. But with all seriousness. So, you know, you mentioned I worked at Coralogix and OverOps, but actually my Code RED moment was a bit prior to that. While I was running and scaling Autodesk's flagship product, AutoCAD. We were building it in Israel, the web and mobile version, and it was a huge success. It was a startup. I was one of the founding engineers. It was acquired by Autodesk. So we hit a massive scale. These were early cloud days. We were running, you know, a Java application. The clients were obviously mobile iPads, iPhones. The backend was on AWS. Unfortunately, there was, you know, it was pre-git. So there's no blame. But someone had inserted a code logic that instead of logging something like an error, it added an entry to the MySQL database. And, you know, it was a Thursday night, which is the equivalent of, you know, don't ship code on Friday. In Israel, it worked Sunday to Thursday. So it was Thursday night. I get a, you know, ping from it that was like a text message from our vendor for uptime saying, you know, I can't reach the APIs. I log in and unfortunately you know, you kind of start to go in and do your firefighting mode, and the database disk was full, 100% full because unfortunately, there was like an infinite loop you know, allegedly. Right at the time, I didn't know there was an infinite loop that just stored events. Right. Analytical events to the database without any you know, consideration of like a, like a circular buffer. And that was it.
Chen Harel: We had to kind of debug the issue, took two hours to, you know, identify and then obviously resolve. Luckily, we were able to, you know, allocate more storage, replicate the database and move it over to a new one. And after a few hours, we got the system live. But it was very unfortunate. For a few hours, you know, folks couldn't really log into their drawings, as you would say in AutoCAD. And yeah. So that's kind of like my code red moment. Because, you know, it was a big fuck up. The production was down for a real application with, you know, hundreds of thousands of users, which is kind of the interesting part of it is that that's the day that we kind of spark the idea of actually building an observability tool because, you know, it's one it's those nights where you're kind of saying, okay, oh, I wish I had a tool that could have just found that infinite loop potential or find that, you know, we have we are where we have a trend that the database is going to run out of steam before we react. You know, we called it the original days for the original name for OverOps. Not many folks know it was actually preemptive software, because we wanted to build something that prevents problems in production again, and just to kind of, you know, for the audience, this is around 2010. So 15 years ago, way, way back, you know, when the, the, the first era of APMs were still kind of lingering. I know you, you, you took part in it as well.
[00:04:35] Chapter 3: OverOps Origins, Funding, and Team
Mirko Novakovic: So I can tell you, I don't know if you know, how I came across OverOps.The first. Time. It was in 2015 and I was fundraising. I was at lightspeed. And I was talking to one of the guys who had invested in app dynamics, John Briones, who then founded Unusual Ventures with Joty, the founder of Harness, who acquired OverOps later. But they wanted to invest in us. And then finally they said, oh, we can't invest in you because we are making this investment in OverOps in Israel, which is kind of competitive. And I was really sad.
Chen Harel: Lightspeed led our series be, you know, it was in 20. It was in 2016, I think, or 2015 when they had our, we had our series B, 15 million, you know, now it you know, some companies are doing that as a seed round, but back then it was a big deal. You know, for an Israeli company, regardless, to raise money in Silicon Valley was a big deal. So we were very happy. You know, the brand was on point. I was very fortunate and very lucky to be, you know, working and building that company. It was, you know, I guess one of the biggest highlights of my career.
Mirko Novakovic: And it also shows that I remember at the time that you got some really good people from dynamics. At the time, I was Steve Burton, I think, who joined it was Randy Boyzon, Randy Boyzon on the sales side.
Chen Harel: The first sales in go to market motion. You know, Nick Durkin, he's a close friend. He's still a running field CTO at Harness. So, you know, a lot of folks came from AppD to OverOps and then moved on to harness with Jody to, you know, to scale that even further. So, you know, some folks have closed the circle with the acquisition. Unfortunately, I was not part of that by then. But, you know, I don't know if you want to do it like chronological or just the highlights, or.
[00:06:35] Chapter 4: Production Debugging: Depth vs. Breadth
Mirko Novakovic: I would talk about the overlaps of the category of OverOps, how you came up with it. And then I would say there are a few tools in that where I would position OverOps is a little bit similar to maybe rook out. Yeah, which was a similar tool. And also Light Run I think today is a similar tool, right? As far as I remember overall. Also highly around getting really the details. So the variables, the, the data inside, it's kind of debugging information in production that you get with that tool.
Chen Harel: Right. You know, it was like the debugging epiphany. Right. You would go into one of those VMs and, you know, this is back when it was still like a largely monolith, right? So there is so much multithreading going. There's this huge context that the runtime itself and the libraries and the frameworks bring. And none of that gets to the logs. Right. And many of the exceptions and many of the errors are hidden, swallowed. So there was a huge pile of use cases for mainly, you know, Java enterprise engineers, .Net engineers that would really benefit from, you know, going as deep as we could in bringing in this context back to like a SaaS experience, right? When you have an error, we used to inject these tiny URLs into log statements so you can click and, you know, deep link into the platform to see, almost like a debugging session without a running debugger. Now, Rook out and Light run came after and took it one step further, even calling it like a dynamic debugger where you get to choose an endpoint like a conditional non breaking breakpoint. That's actually a feature we used to have in the platform. But just because of the nature of the JVM at the time the stability of dynamic instrumentation in a very ugly production environment just felt too sensitive for our customers and our kind of ideal experience. So we actually sacked that feature, which then grew to be like a category of itself, even Datadog. Today they acquired the company. They have their own dynamic instrumentation or dynamic debugging product. Yeah, I can say that I've built some of that in C plus. Plus, you know, in 2013 but we never materialized it. We never stabilized it enough to be worthy. Yeah. And I mean.
Mirko Novakovic: I have to say, because we also developed agents and we looked into this. It is a really hard technical problem to solve on the, on the agent side. Right. Especially because you need to intercept an application really deeply. Get a lot of data out and, and that without really disrupting the production environment or adding so much overhead that it gets.
Chen Harel: The industry wanted the three, you know, 3% max CPU or latency overhead to the transactions. You know, OverOps had amazing customers, you know, JP Morgan Chase, Two Sigma. British Telecom, out of the UK. Comcast. So like big enterprises with huge problems but ultimately. Right. This you know, I remember we met in, in our headquarters in San Francisco and you kind of you kind of mentioned like Instana play of like having, you know tailored engineering experiences for the different you know, runtimes. And here I am telling you that I have like, ten x intelligence force hackers that know all the bits and bytes of the JVM, like we're the best at Java runtime. However, the market evolved, right, with Node and Python. If you are not polyglot, it's very hard to actually make, you know, move the needle. I remember at the time we looked at sentry as like the complete opposite, right. The simplest, the simplest tool. But they support everything and they just, you know, give you something on everything. But we couldn't really pivot fast enough to, to give that, you know, breadth rather than depth.
[00:11:05] Chapter 5: Market Realities and Scaling Challenges
Mirko Novakovic: Well that makes sense. And overall I have to say. I mean. Correct me if I'm wrong. I mean, Lightrun is doing pretty well. I think they raised a large 70 million round recently, but all the other tools got acquired at some point. Pretty, I would say pretty early in the I mean, at least on, on a revenue scale, they never really scaled out to become a large company, I would say.
Chen Harel: So first of all, I completely agree. And I think, you know, you and I are yet again, in some capacity working in industry. Right? And there's just cycles, right. Can you build, like, a 20, $2,030 million company. Can you build a $200 million company? And all of us look like a Datadog or Splunk, and we want to be like a multi-billion dollar company? New Relic was, you know, a decent product. Sumo logic is a decent, somewhat product in its own category, but, you know, they couldn't really make it. So there's, there's like just with everything. There's a pyramid, in my opinion, and it's very hard to really move the needle on top. Because of the requirement of the, you know, technical engineer or the DevOps, the SREs, they just have a lot of demands. It's hard to execute in that space.
Mirko Novakovic: Oh, absolutely. I totally agree. I have to say New Relic made a great job, but I think they got to a billion in revenue, right? Which is awesome. And they were one of the early cloud APM vendors who really made it. Yeah, the first SaaS product in that space. Right. So, Lucerne and the team did a great job. And but you're right I mean, both Sumo and New Relic got acquired by the same private equity fund.
Chen Harel: So it means someone there as well. Like I would say, gave up maybe on, like, you know, we can beat data though.
Mirko Novakovic: Yeah, absolutely.
Chen Harel: I don't know how we know, we haven't really talked about what's your take on, you know, or are you in the game of beating Datadog this time around. Like how do you think about it?
[00:13:04] Chapter 6: Competing with Datadog and Market Cycles
Mirko Novakovic: I like that question on this podcast. I mean, I have to say absolutely, right. I mean, my goal is to, I mean, what does beat mean, right? What I tell my folks all the time is this is probably, it's hard to say at 20 to $40 billion market. Right. That's where we are. Datadog is yeah, the leader in this space. They have an amazing product, amazing execution all. I mean, I have all the respect for the team there and what they've built, but they have probably something between 10 and 10 and 20% of the market. It's not a Salesforce type of category where you own 7,075% of the market. I think it has never been the case. And the reason for that is, as you have said, the market is changing all the time. I think there are cycles of you can say I sometimes say seven years, sometimes ten years. If you look back to Wiley in 2000, the first kind of Java agent, then New Relic app dynamics, Dynatrace, which, by the way, Dynatrace is one of the company they have reinvented themselves inside of the company, I think twice already and others don't.
Mirko Novakovic: For us, it would take a few years to be able to compete with Datadog on large scale deals, but at the time, I think we as a startup, we are built on top of OpenTelemetry. We are natively built around the standard. We are easy to use. We use a modern UI. I think we use AI pretty well inside of the product, so I think we have a chance to beat them over the next few years. But then I also have to say Datadog will probably, at that point in time, be a, I don't know, maybe $10 billion company also in security, built into really large deals into large customers. And they will, I mean look at Oracle right. They do. Amazing. And you will, I have already said 20 years ago that nobody needs that database anymore. And they are still around with a lot of products. So I would say beating is hard work, right? We think.
Mirko Novakovic: We will find a place in that market to become really big. I would say it that way.
[00:15:21] Chapter 7: Global Roots and Differentiation Strategy
Chen Harel: No, no. So first of all, you know, I always like to acknowledge when I speak with folks about observability that you know, Datadog. Dynatrace. The companies I work from, from Israel. Right. There's something in European or, you know, like non US companies in this space. That's really interesting, right? Where engineering is outside of the US and, you know, innovation and finding new strategies to, to bring to market is super interesting. And obviously data that we're first with like cloud monitoring Dynatrace has an amazing APM holistic solution. And then, you know, companies like Coralogix that I worked for, you know, there are even I guess like newer generation Dash0. Of course there's ground cover out of Tel Aviv, more ebpf focus. They're really trying to disrupt how things were made.
Mirko Novakovic: Also, Grafana, by the way, Swedish Torkel works. So yeah, of course a lot of the companies have European Israeli roots. Yes.
Chen Harel: No, no, for sure. And you know Ariel who I know he's also a friend of yours. You know, he always say different is better than better, right. So like, as an entrepreneur or someone who's kind of like trying to to fight an incumbent. Always find the angle for what you do. Different because different matters. You know, you just have to convince yourself and obviously execute especially.
Mirko Novakovic: I talked with other I mean that's a very important thing for founders, right. It's especially true in a large market. I mean, this space is not a blue ocean market, right? There are a lot of big vendors with multi-billion dollar revenues, right. Like we mentioned, Splunk, Dynatrace, New Relic, Datadog Grafana. I mean, there's many elastic, I mean, a lot of also European roots, by the way.
[00:17:14] Chapter 8: Pricing, Paradigms, and the Miro Win
Chen Harel: I is a close friend. You know even though he was building it from Amsterdam, you know, he's Israeli at heart.
Mirko Novakovic: I totally agree. You have to find your niche. You have to start in a niche because all the vendors are also Coralogix. Right? They started logging in, in a very special segment. And then from there you grow into an observability platform. And then probably, as you see at the moment, you then also outgrow that market and you add an attempt to that by adding I mean you can see with Datadog right there added security. I just read they are acquiring
Chen Harel: Wine. Yeah.
Mirko Novakovic: Israeli and company for $1 billion. Right. Cybersecurity company.
Chen Harel: Amazing venture there. You know, great, great guys. I know them for sure. You know, the security play is always at heart, right? I don't remember if with Instana you had like a, like a security play. I don't believe so because let's focus logs. But you know, Coralogix for years we have a like a speculative venture called Snow Bit with managed services and features built for security. I'm pretty sure that's like definitely more than half of Splunk revenue is tied to their ITSM and security. It is a big deal, right? My CISO used to say that all events are security events, right? So, as long as you, you know, compliance reasons, the security team, they have to store them, maybe even index all the data. So, you know, what better tool than an observability tool would be to crunch the numbers and give you insights. Who knows what Cisco is up to now? Right. They just like to acquire observability tools. And I don't know if there's, you know, we'll see what happens.
Mirko Novakovic: It's like a collection, right? Every few years, they acquire a new one.
Chen Harel: Well, well, here, here in Israel, we're still kind of baffled with the whole Hebsagon acquisition.
Mirko Novakovic: Oh, yeah. That was amazing, right?
Chen Harel: That was, you know, those founders, Nitzan and Rahn, were just like, you know, throw that. Just throw money at them. And then what happened? Like, I don't know, it's just. But this is getting this is me getting, like, old and grumpy and, you know, jealous, right? So I apologize. It's just amazing.
Mirko Novakovic: No, no. It's good. I mean, at the time, it was shortly after the Instana acquisition, and I have to say Epsilon was less than 10% of our size and they got the same price. They got the same price, literally. So I have to say take my hat off that they did it right.
Chen Harel: Then and they never integrated like we know because, you know, we helped you know, companies migrate away from the epsilon, you know end of life product. But there was nothing to migrate to like, just just I don't know, it was I don't know if it was a political battle. You know, obviously I wasn't inside.
Mirko Novakovic: I have some inside information I can share here. Right. Without naming, but what I heard is that there was really a political battle between the especially the dynamics, the FT guys, and they didn't want to integrate it. And so then they put it into the security team. So Epsilon was put into the security team where it made almost no sense. And then the product died. Right? I think part of it became open source.
Chen Harel: And yeah, it's ridiculous and you know, by the way, I have some, you know, in sake of, you know, some, some inside information. Right. I'm sure you have your own stories from the IBM acquisition. At Coralogix, we partnered with IBM on the whole Cloud Logs product. We always kind of waited for the Instana guys to kind of enter the room and, and consult, but it's so hard in these organizations and all the VP's or the GM's are trying to keep their cards close to their chest. There's a reason why I do startups. Like, I don't know how to play that game.
Mirko Novakovic: I mean, I was only briefly at IBM, but I talked to the cloud team at that time. I think they used New Relic and for the cloud, and we tried to position Instana and it almost feels like an external customer, right? If it's a different division, you sell to an external customer internally. And you also have the feeling I have no benefits. It's the other way around. Somehow, because you are inside of the IBM, you even have a disadvantage to sell. So I yes, I know that and congrats.
Chen Harel: I mean it's a great partnership. It's an amazing deal for both companies, right. They wanted to, you know, get rid of their previous vendor. They, you know, try to build something their own and realized this is also a bit of a big, you know, rock to, to move. And they partnered with a vendor that's hungry and is willing to do the effort to to be like a first citizen you know, cloud product for them, you know, white labeled all, you know, all the way.
Mirko Novakovic: And talking about CoreLogic where you have been VP of product. I think one of their, as far as I see it. Right. Also sometimes in deals it's really competing also on cost. Right. So one of the key benefits is... By the way, at an Excel event I talked to one of the founders of Miro, and I know that you replaced Datadog there, right?
Chen Harel: We really did.
Mirko Novakovic: We did probably a pretty big deal. And also, I think there was even a Datadog person on the board of Miro. So it was not easy politically to get that replaced. But you did it. It is probably also because of a pricing topic.
Chen Harel: I'll briefly touch on that, you know, trying to stay within reason of what's you know, can go on the public record. But, you know, internally, we are really focusing in the DNA of the company is to not be cheaper. Like, it's not something that we put and we try to advertise this is a, you know, a cheaper product. However, it is less expensive, so it's very hard to not discuss it, especially when you see that the number one pain in this space, people coming from Splunk and Datadog as the market leaders. Let's give them 30% combined of the market, right? A lot of the top companies are using those tools. They want to reduce costs. The vendors, let's call it the legacy vendors, just don't have it in them to offer, you know, a better or a different way. So what Coralogix did for years now is to just bring a different paradigm, a much simpler pricing model, and a simpler architecture to think about data, right? As opposed to a long list of SKUs and discounts and computers. You know that you need a PhD to even understand how to read like a Datadog usage or a bill. Right? Kind of similar to AWS to some, you know, to some degree, it's too transparent to the fact that you just can't really grasp it without like an LLM Coralogix just simplified it.
Chen Harel: Yes, it tends to be, you know, 50% cheaper, which is great. But I have to say that this is like, not this is not the strategic play because there will be another vendor because, you know, Coralogix now is like 500 plus people. It's, it's it's own beast now. So there will be another company, let's just call it, you know one one that will say, hey, we're 50% cheaper than CoreLogic. So, you know, why don't we do the transition again and then, you know, in a few years, there will be Dash0 that say, hey, you know, as a startup and you know this, right? You are, you get in the beginning, you can be much more creative with how you get contracts. So, you know, Coralogix we were really selling the value of, you know, the paradigm shift in architecture where the customer gets to keep their data. You know, it's not as crazy in open source, you know, all only open source like Grafana would. Because Coralogix does have some proprietary aspects to it.
[00:25:35] Chapter 9: From ELK to ClickHouse and New Architectures
Chen Harel: We do analyze the data on our end, and we pay for and we pay and we do the compute. But all the storage in open formats gets to stay on customer's premise for deep analytics for, you know, obviously, you know, beyond Coralogix, it's very easy to make it away. And I'm not even talking about like, the whole pipeline of getting data in which, you know, it's own conversation. Actually, it's it's own category today in the observability space of what, what we just call pipelines now. But Coralogix did this amazing split between compute and storage. And even then, even there, I see folks taking it even one step further. Right. Companies that are saying, hey, just, you know, put everything on your premise will do. Like a managed cloud offering. Will we, you know, give us a service account, will deploy on prem on your premise and we'll charge you, you know, you'll send telemetry out, we'll charge you for the, you know, amount of nodes there. You have the amount of data. So there's even there when, you know, Coralogix made one step, there is a company that will make another step. And it's interesting.
Mirko Novakovic: Absolutely. And you can see that there are companies like Axiom, for example, right, who take that a step further.
Chen Harel: If I did and, you know, if you think about it, one last thought on the matter, right. Coralogix started, as you know, putting some machine learning and managed elk right back in the time there was elastic. But even the cloud elastic was not serverless, right? You had to think about servers. You had to think about disk space, decide and know how to be like a DBA of sorts.
Mirko Novakovic: Re indexing and stuffing.
Chen Harel: Exactly. And Coralogix and other companies in that space said, no, we want to you know, we we will build like a SaaS multi-tenant solution on top of that, which for a few years made sense. But now even Coralogix all the companies need to be like again differentiated. And you know, elastic was great. And they have a very interesting story with, you know, with the vector DB portion of an LMS. But, you know, my eyes are on ClickHouse and how they're building and positioning themselves to be the next big thing in our space. And, you know, because a lot of vendors are building on click house clickers themselves, acquire the company to have a coherent observability story as an offering.
Mirko Novakovic: HyperDX. Yeah. HyperDX.
Chen Harel: Yeah, I would definitely buy some stock on the secondary market if I had the money I, I, I really I really like how these guys execute.
Mirko Novakovic: Absolutely. I mean we are in regular discussions with them as we are a big user. We already, by the way, we used it in 2016 when it was still a Yandex project and open source, and most of the documentation was in Russian, and we found it because our CTO Pavlo was Russian and we started using it. We were amazed by the scale and the speed of that database. Right. And so we used it early. So we are big fans for almost ten years. Amazing product. Absolutely. And you especially see that there's a nice story of Shopify, how they build their own observability stack on top of Clickhouse. Highly scaled.
Chen Harel: And I think also OpenAI just released some paper on their own observability stack like there there's definitely and this is also what I mentioned earlier about, you know, the whole you know, I don't know if it's private or, you know, the whole private cloud experience. By the way, Coralogix had some private cloud clients as well, where we literally installed the entire Coralogix platform. And I'm saying entirely because it's a big platform, it's a big microservice architecture of Kafka Streams. But newer or I would say newer projects are essentially doing like a web app in front of like a click house. Nice UI and let's call it production, right? It's like three services, very simple architecture, very easy to embed or to install on premises. So there is merit in that as well. If you can get, you know, all your checks in order.
[00:29:55] Chapter 10: DIY at Scale: OpenAI and Shopify
Mirko Novakovic: Absolutely. And at the end, I mean, if you look at Shopify or OpenAI, I mean OpenAI, I think recently I saw a downgrade by an analyst of Datadog because OpenAI pays data like $170 million a year based on that article, which is 5% of the overall revenue. And because they published this, building their own observability stack. There was kind of a risk of saying, okay, what if they churn? And then.
Chen Harel: But they will. But they will turn.
Mirko Novakovic: A Shopify, right. Shopify I think, paid 30 million. There's this presentation around it, 30 million a year to Datadog. And now they have 50 people building it on top of clickers. And I think they said overall it costs 7 or 10 million.
Chen Harel: There's an analogy I would use, right? When you and I owe the bank, like, you know, $1,000, the bank is angry at you. But when someone owes like a hundred million to the bank, the bank is their friend. Yeah, right. So, like, there's definitely a dynamic or like a, like a power shift here. Where, where OpenAI now or companies that are paying, you know, getting to like the tens of millions that they can really call the shots. Even at a company that has like a 50 billion market cap. You know, I remember at Coralogix when we saw that some of our customers can get too big. This is where we went to the private VPC. We wanted them to be very sure that there is really no reason or no way, like no realistic way, that they can get it cheaper by building it themselves, right? This was strategically crafted. Otherwise, you're at a risk of, you know, one of those turns to essentially kill your next round at a startup.
Mirko Novakovic: That's true. But I also have to say that at a certain point, it makes sense to build your own tool at the scale of OpenAI or Shopify or others, because then you could be much more opinionated, because if you want scale in performance, then at some point you can have a multi-purpose observability tool. You have to make certain choices, right? Because, you know, these are my attributes, then you can index differently. You can optimize the database differently, but you can't allow an okay query for everything in every time frame. Right. That's and if you look at Shopify, they did that right. They made some pretty good choices and were very opinionated about their stack. And then they can build something that is really scalable and super fast for them. Right. So I totally see the point why you would do that at a multi-million dollar scale.
[00:32:37] Chapter 11: Multi-Vendor Reality and a New Interface Layer
Chen Harel: My take there would be, which is kind of interesting, right? We see enterprises and OpenAI, even though they're an enterprise in scale, they're still very young at heart. Right? We talk about those born in cloud companies. Obviously, OpenAI is even like a in a different sphere in terms of like the, the cutting edge technologies that they can acquire. But, you know, if you take some enterprises, we are still baffled that enterprises just have like a, you know, they have all the vendors, right? They pay Splunk and Datadog and Coralogix may be. And they don't even anticipate a consolidation. Play so soon? Right. Like, whether it's different business units or just, you know, for different reasons, just vendors stick around. It's very hard. Again, it's hard for me to name drop here, but I'm sure you know it. Right? Like when we have enterprises that are just paying multiple vendors, which kind of makes me think whether there's a play there as well. For some, I don't know, let's call it like an AI or some consolidation that's happening from the outside. Right. Rather than trying one of the vendors to eat all the rest. You know, maybe one of the new companies that are doing AI SRE or something in that realm to just say, hey, we don't care who the vendors are, we are the tool that is going to be your next interface.
Mirko Novakovic: Yeah, that's a good point, right? We are also thinking about that also. We think it could be that one of the coding agents becomes the interface, right? You integrate your observability tool into the cursor. Yeah, we are an MCP server. We do that. And then you could prompt and say, hey look at the issuing, Dash0 and then suggest me a fix in the code. Right. So then you don't even have to look at the tool. So it is, it is, it is hard at the moment to predict how this whole thing will be.
[00:34:36] Chapter 12: AI Frenzy, Acquisitions, and UX Lag
Chen Harel: And luckily I'm not a VC, but like I we we can we can we can play the guess, you know the guessing game. But things are moving at such a crazy speed. You know, we talked a lot about startups, right? If you see what's happening in that realm with acquisitions, reverse acquisitions, the acquisitions, I'm reading the news and I'm like, how is this legal? Of all the things that's happening there is.
Mirko Novakovic: It's not really understandable. Right? Yeah.
Chen Harel: I'm like, I'm like, why didn't they stay in college for a few more semesters of, you know, machine learning?
Mirko Novakovic: Yeah. Windsurfing is a pretty crazy deal, right? Or yeah. Scale AI and whatever happened. Right. That's really large numbers, but that just shows that those companies who are acquiring them are so big in the trillions of market cap and I am threatening their business model. And so the amount of money doing their best. Yeah, the amount of money you can invest to reduce the risk is just so high, right?
Chen Harel: I have to say, I'm still so disappointed that the interfaces that I use daily, Siri and Alexa are still so far behind. It's just crazy. It's crazy. How they're just like, you know, years now behind what you and I are doing with cursor, for instance, or ChatGPT.
Mirko Novakovic: Of course, or just use. Chatgpt. Right. I use this in my car all the time, right? Talking to ChatGPT, taking notes, getting some stuff.
Chen Harel: I'm telling Siri to go to get ChatGPT on the phone. It's ridiculous for me, you know, a user experience from that in standpoint.
[00:36:19] Chapter 13: Closing Reflections
Mirko Novakovic: It's fun talking to someone in this space who knows the whole history and what's there. It's almost yeah. More than ten years. Right?
Chen Harel: Yeah, I'm telling 15. I've done engineering product solutions, but I'm a builder at heart. So you know dev tools are like My space. And yeah, I appreciate the invite. I enjoy talking with you again. Mirko. You know, the next time I'm here.
Mirko Novakovic: Thanks for listening. I'm always sharing new insights and insider knowledge about observability on LinkedIn. You can follow me there for more. The podcast is produced by Dash0. We make observability easy for every developer.
