In this episode of Founded & Funded, Madrona Investor Rolanda Fu is joined by Dedy Kredo, the co-founder and chief product officer of Qodo. Formerly CodiumAI, a 2024 IA40 winner, and one of the most exciting AI companies shaping the future of software development. Dedy and his co-founder, Itamar, are entrepreneurs who have spent their careers building for developers, and with Qodo, they’re tackling one of the most frustrating problems in software engineering — testing and verifying code.
As AI generates more code, the challenge shifts to ensuring quality, maintaining standards, and managing complexity across the entire software development lifecycle. In this conversation, Dedy and Rolanda talk about how Qodo’s agentic architecture and deep code-based understanding are helping enterprises leverage AI speed while ensuring code integrity and governance.
They get into what it takes to build enterprise-ready AI platforms, the strategy behind scaling from a developer-first approach to major enterprise partnerships, and how AI agents might reshape software engineering teams altogether.
Listen on Spotify, Apple, and Amazon | Watch on YouTube.
This transcript was automatically generated and edited for clarity.
Rolanda: Well, before we dive into Qodo, Dedy, could you just share a little bit more about your journey? You’ve navigated diverse roles across the tech landscape. You and your co-founder both are still entrepreneurs. What pain points did you guys experience, a developer workflow that made you say, “We have to solve this, and AI is finally ready?”
Dedy: So I have diverse background. I’ve been in engineering roles, data science roles, product roles across both small startups and larger organizations. And I think throughout all of this, software quality has always been kind of near and dear to my heart.
It’s always kind of a challenge to strike this balance between wanting to move fast, develop fast and providing high-quality software. As a startup, you’ve got to be moving really fast. And I think with AI, it’s now becoming even more important now that the market is changing really, really fast. It doesn’t matter which field you’re in. If you’re impacted by AI, you have to be moving really fast. But you have to strike the balance with also providing high-quality software, and that’s always been a challenge.
So Itamar and I have known each other for many, many years. And basically, we realized that as AI starts to generate more and more of our code, the challenge kind of shifts to how do I make sure that my code is well tested, well reviewed, secure, aligned with the company best practices, especially in very, very large enterprise organizations?
That was a challenge that we felt that is going to become the next frontier. And we realized that pretty early on. So if you look at our seed investment deck from 2022, it’s kind of like we’ve been seeing the same pitch for a while now, and I think now it’s actually all coming together where we’re really well-positioned for this. So yeah, it’s exciting times.
Rolanda: So you and Itamar are both serial entrepreneurs and have known each other for a long time. How do you think about whether you were the right founders for building this? How do you think about this new category of intelligent coding platforms, right? How did you think about this idea, and how did you get the confidence to build in this space?
Dedy: Yeah. Well, on one hand, for both of us, software quality has always been very near and dear to our hearts. I have a lot of experience working with US companies, with large enterprises. I spent a lot of time working with financial institutions, for example. And Itamar was a CTO several times before, and he was kind of ready to step up for the CEO role, and we share a lot of the same values.
We grew up in the same small town halfway between Tel Aviv and Jerusalem. And then, yeah, we just knew that this is very interesting times and very exciting times and that basically software engineering is being reinvented and is being transformed in a massive, massive way. And we believe that the right way to enter or penetrate into this market is through enabling organizations to embrace AI for software engineering in a responsible way. And we’ve had a similar pitch since day one. Yeah.
Rolanda: Yeah, that’s awesome. And maybe with that, let’s dive into the Qodo platform a little bit more. What’s been your guys’ North Star the whole time since those early seed days, and how do you think about what you have versus the broader space of the plethora of AI coding tools out there these days?
Dedy: Basically, there’s a lot of excitement, and I would say some hype also around AI code generation, a lot of talk around vibe coding, and how AI is going to write everything. And we believe that for enterprise to really embrace gen AI and really have gen AI impact their organization in a way that helps them to increase productivity significantly, they’ll have to find a way to kind of balance this with quality and with making sure that code is aligned with the best practices, that code is well tested, well reviewed.
And in order to do that, the foundation for everything has to be very deep understanding of the enterprise code base. So this is something we’ve been investing in a lot. The foundation of our product is something called Qodo Aware. It’s a layer of understanding code bases, indexing code bases, and understanding how different components relate to each other in very large code bases. So that’s one major area of focus. And then on top of that, we have two major product areas. One is around code review and code verification.
So this is our Qodo Merge product that integrates with different Git providers and basically helps take code review to the next level. Because if you think about code review, it hasn’t changed in decades. And basically, developers open a pull request and they start reviewing diff by diff and trying to figure out if there are issues or bugs or anything like that. And with Qodo Merge, we make pull requests a lot less painful. We help developers understand the actual changes, and we create a detailed logical walkthrough.
And then we also try to catch bugs, and we automatically generate best practices for each team and each repo and for the organization as a whole. So that’s on the code review side. And basically, we believe that as more and more code gets generated by AI, the bottleneck shifts to how do I review this at scale? How do I maybe auto-approve code that is maybe smaller changes or that doesn’t have any major issues, but how do I help developers review code at scale and catch issues fast?
And then we have the code generation side, where we basically have different IDE plugins for different… for various IDEs. Our approach is basically not to have developers need to switch their IDE into something different. So we integrate with existing heterogeneous environments, both JetBrains IDEs, for example, as well as VS Code. And then we’re just about to launch our CLI. So essentially, we have the same kind of coding agent in the background that is driving both the IDE plugins and CLI, and also agents that would run in the background.
And all of that has in the back of its head, you can think about it like that, the coding agent, has the company knowledge and best practices, and that’s kind of what unites us… unifies us. So basically, we look at the SDLC in a holistic way. And then one last thing I would add is that we believe that we have a strong belief that developers and enterprise organizations will need to customize AI agents for their specific needs.
So we don’t believe in a one agent that would rule them all. We believe in more of an agent swarm type approach where different teams will configure agents a bit differently, will give them different tools, maybe different permissions, and would want to control the input and the output, and the triggering of the agents. And we built a system to enable them to do that.
Rolanda: But I think that’s one thing that I really love about your guys’ approach is, right, kind of that end-to-end development lifecycle coverage, whereas I think a lot of tools out there tend to kind of pick an area to focus on. So I think that’s really clever on your guys’ end.
I am a little bit more curious, too, to dive into the platform. I mean, it seems like you guys have built a lot. Can you talk a little bit more about that decision to do a lot of that development versus leveraging existing models out there? How do you make those trade-offs?
Dedy: Generally, we do have… we do leverage the large models, and you have the ability to choose the companies that use our product, leverage both Anthropic, OpenAI, Google models. So we also have very flexible deployment options. You can use our SaaS. We also support single-tenant environments and self-hosted. For self-hosted environments, we do provide our own model that is essentially built on top of an open-source model.
So we don’t train a foundation model from scratch, but we did invest quite a bit in training embedding models for code because we believe that the foundation for… as I mentioned earlier, the foundation for everything is deep code-based understanding, and we saw that there was a gap in the market in that area. So we did train a state-of-the-art embedding model for code that has comes as a built-in part of our platform.
Rolanda: I love how dynamic you have set it up to be, and I think that’s really critical for scaling any kind of solution these days. And maybe just to pivot a little bit, I think a topic that’s really on everyone’s mind these days is this term around vibe coding. So I’d be curious to get your guys’ thoughts too. How does your platform enable the vibe coders of this generation to better leverage your platform, and how does that impact what you have created?
Dedy: Vibe coding, I think, it’s like… And I think also when Andrej Karpathy coined that term, he was really referring to pet projects for coding when you don’t care about how the actual code is being built. You’re more focused on the functionality and just seeing that a functionality actually works, but that’s not sustainable for enterprise production code. You’re basically generating a lot of tech debt, and you may be overlooking issues. You’re not focused on testing. So, in order to make this process of AI generating the code work for these complex code bases, you’ve got to put the right processes and tools in place that allow you to check the code to set the right frameworks and best practices.
So that, first of all, you try to get the code to, right as it gets generated, already get to take into account your different rule and best practices for a given code base for a given team. So we do that with our Qodo Gen with our generation side.
But then also once you need to review the code, that’s the checkpoint, that’s the point where you got to really make sure that it’s aligned with the best practices, that it’s well-tested, well-reviewed. And we believe that having these two sides work hand in hand, we call it the blue team and the red team, and that makes it actually work in an enterprise environment.
Rolanda: I think that’s a really good description of both sides. The red team, blue team have to play a little bit of both. And I think that’s something we’ve talked a lot about internally, too, is just people talk about code generation a lot, but not enough about some of the other sides around testing and review. And I think those are even more critical it seems in this current environment, especially with something like vibe coding.
So I’m curious, you mentioned enterprises. Can you talk a little bit more about how you balance getting developer love versus selling to enterprises? Is it one or the other, or is it a little bit of both? Have there been any pitfalls when trying to focus on one or the other, or has it always been a smooth ride?
Dedy: It’s definitely a challenge. Generally, with a startup, we try to focus, and for us, we call this strategy middle-out. Our focus really resonates with team leads with architects, with platform teams, and developer experience teams, which by the way, you’re now seeing these kinds of teams at large organizations gain a lot of, I would say, power in the organization or ability to influence the tools. So we are assisting these teams to really grow.
So on one hand, our pitch really resonates with higher-level managers, architects, and team leads, but on the other hand, as a dev tool company, you ahve to have this bottom-up approach, and developers need to love using your product. We’re trying to always balance that. So we go both top down and bottom up. And it means that we do have a self-serve approach, and we have a freemium tier. We do have the ability to swipe a credit card and go to our team’s offering.
But typically, when you get to the enterprise side and you want to index a very large code base and you want to do it in a single-tenant secure environment, that’s where you do a more controlled proof of value and you engage in the conversation with the enterprise stakeholders. So a lot of our, I would say, larger customers, they tried us out self-serve, they just came and experimented a little bit with our product, but then they contacted us to do kind of a larger trial or a larger pilot. So this is how things have worked for us generally.
Rolanda: That makes a ton of sense. I think the balance of both is super critical, both the individual and the enterprise level. I’m curious, are there any stories of enterprises that were hard for you to get that you were probably most proudest of converting, or any horror stories, I guess, of trying to sell in the space, or even advice for people that are trying to sell in this space?
Dedy: I can give an example of a Fortune 10 retailer that is really one of our largest customers, and they really like… their challenge was around, “How do I make sure that the code that gets generated by AI is well-tested and well-reviewed? And how do I kind of…” Their focus was a lot on the code review bottleneck, and they approached us and they started small with a small pilot, and what they saw is that the product just started start expanding in organization. People were hearing about it and were wanting to turn it on in their repos.
And the challenge that we had was really around supporting the growth that they had inside of their organization. All of a sudden, they had thousands of developers knocking on the door, and this is an air-gapped environment. So you have to take into account things like load on GPUs and things like that, making sure that response times are good and that the quality of the results are good and that it’s aligned with their best practices of the different teams.
So we worked with them very closely to be able to support them, but yeah, they expanded and now they standardized for the entire organization on Qodo for their entire pull request and code review process. So yeah, it was a journey. So the pilot was a few months, and then they expanded, and it took time until they expanded the entire company. But yeah, it’s like with these companies, you’ve got to really support them. You have to really give them the feeling that, or not just giving them feeling, but really work very closely with them and listen to their pains and be willing to kind of go the extra mile for them.
Rolanda: That’s kind of the dream land-and-expand scenario with a customer, right? Hopefully, they’ll be your customers for a long time coming. I’m curious, I mean, given you have spent so much time with these developers and these enterprises, how do you see the future of some of these developer teams changing? Are you already seeing how Qodo is impacting how these teams are structured? I’m curious where you see the future of all this going? Are there… Are engineers going to all be replaced? I think that’s what everyone’s scared of, right?
Dedy: The way I think about it is that I think that the roles of developers are just changing. I think that, especially for very large complex code bases, I don’t imagine a world where a product manager can in a click of a button just make a change that impacts the entire code base and redo the entire onboarding experience for a new customer for one of these very large. Maybe a large bank or any kind of large enterprise. So I think the developers are going to become orchestrators of agents. Each one of them will have the ability to launch multiple agents and also customize each agent for specific use cases, specific triggers, and then be able to review the work of these agents at scale.
And then, yeah, most of the code, they’re not going to hand write, I would say. But they need… I think there’s still going to be, at least the way I see it for the foreseeable future, for complex code bases, you’re going to need technical people, technical developers experience that are able to orchestrate this work and make the dev teams a lot more productive, but also make sure that you don’t have this kind of, I hope it’s okay that I’ll say it, but a CrowdStrike moment where the world grinds to a halt because something was overlooked. So the way I see our goal as a company, or what we’re trying to do, is enable these organizations to be so much more productive, but not have these CrowdStrike moments.
Rolanda: And is this something that you see playing out, I guess, over the next five to 10 years? I mean, I think you talked a little bit about the near term, right? I think that makes a lot of sense with the developer kind of as the orchestrator. How do you see this playing out even further out? Is there even going to be an entry developer role, or how do you see Qodo really being that partner in terms of really catalyzing this change?
Dedy: First of all, it’s very hard to make very long-term predictions. But the way I see it is that I think the role is going to continue to evolve. I think we still need… There may be some kind of a curve in the demand for developers where you see maybe going down and then going back up because you’re going to need these people that are very, very technical, that are able to orchestrate and manage these agents that are writing code.
And if fewer people end up now going into studying computer science and things like that, then you’re going to have a situation where you don’t have enough of these people. So I think we’ll see that there will be very interesting dynamics. Also, I think there’s going to be an explosion of software in general.
Think about all the ideas in people’s minds that for software that are not becoming companies today. I think there’s still so much more potential for a lot more software to be created, and you’re going to need engineers for that. So yeah, I do believe that engineers are going to be mostly orchestrating agents in the next 2, 5, 10 years. But I think you’re still going to need the engineering team for the foreseeable future. This is how I see it.
Rolanda: I think that’s a great assurance for any engineers listening in on the podcast. And I think it’s also something that we’re excited as investors, right, and something that we believe in, as just a lot of these forces will continue to multiply, and more software just means more things for people to manage. I think it’s more about the roles shifting. So totally aligned there.
So maybe just to pivot and switch gears a little bit, I think one thing that’s impressive is just around how fast you guys are growing. So I’d love to hear a little bit more about how you think about go-to-market. How do you make sure that you’re targeting the right customers and training your reps?
Dedy: It’s funny, we’re just now doing an onboarding bootcamp because we significantly grew the team. We’re going to be 80 soon in the company.
Rolanda: Wow.
Dedy: Yeah, I think maybe a year ago we were 30 or so, something like that. So, we’re actually now experiencing this growth, and how do you do that? I think you need to, first of all, do this… spend a lot of time as founders, and we were like 10 people in the founding team, something like that, when we just started [inaudible 00:23:10] company. So you have to have the founders and the founding team really working closely with the go-to-market team, helping them, supporting them, joining them on calls, and make sure that you’re constantly enabling them.
You’re constantly over-communicating. Also, as the market shifts or the market… there are changes in the market, you make product decisions, and you got to make sure that people understand where your product is also heading, and do a lot of product roadmap sessions with both, actually, the go-to-market people, but also with your customers. I think it’s just spending time and making sure you do that.
Rolanda: Yeah, yeah. I mean, I think your job’s only going to get more exciting and harder now to scale out the team and transition from that founder-led sales. But yeah. No, I’m sure you’ll do a great job there. Maybe going off that, I’m curious for your guys’ founding journey so far.
What kind of advice do you have for other people building in the developer space and in AI in general? Are there any hard-earned lessons that you have come across in the first couple of years that you would like to share with some other people that are starting to embark on this journey?
Dedy: The challenge with this AI space and AI and software engineering also now, there’s so much noise, so much going on, and you have to have an insight. You have to stick with what you believe, and you have to find the right balance between building to the future.
So, building for where you believe the models will be in X time from now, but you can’t build for too much into the future because… so you have to strike that balance, right? You need to… On one hand, I would say the balance is probably building for a few months out, where you believe the model capabilities will be, and then just stick with your insight. And yeah, it’s like you either win big or you fail big. So I don’t think there is an in-between at the moment.
Rolanda: Yeah, that’s great advice. And I’m curious, how do you maintain your own long-term focus and integrate customer feedback, right? I think a lot of founders struggle between, there’s a lot of noise in the market that you hear from competitors, from probably your investors, from different kinds of customers. How do you maintain focus between you and Itamar to make sure that you continue to build for that right three-month direction?
Dedy: On one hand, you have to stay up to date. You can’t ignore the competition. You have to strike a balance — on one hand, you do need to react to things that are happening. So if all of a sudden there’s a new model that comes out that allows you to do things that you couldn’t do before, you do need to respond to it, but you can’t just be reactive.
So you need to have a roadmap, you need to stick with that roadmap, but then you also need to build the organization in a way that people embrace the change. I think the people that are best suited for fast-growing startups in the AI space are those who, on one hand, have this kind of grit that can stick with things. And there are hardships. There’s moments where, all of a sudden, we worked on something and some competitor released it a week before we were about to release it, and it kind of sucks this situation.
So you need to have people who can deal with this kind of situation. But on one hand, you have to be very, very adaptable. So you do need to see what’s going on in the market and be determined. I mentioned earlier that one core company value is no fear of good conflict.
We always debate things even between us as founders and with the founding team and with a broader team, but also move fast with confidence. So once you decide on something, you have to move on it fast, and you have to do it with confidence, then you have to make decisions. The biggest issues happen, I feel like, when you get pulled in different directions and you end up not making a decision.
Rolanda: That makes a ton of sense. I mean, this has been so incredibly insightful, Dedy. I just have a few rapid-fire questions for you to wrap this all up. I think we’ve agreed on a lot of things so far. So maybe just to spice it up a little bit, the first question is, what do you believe about the future of AI and software development that many people might not fully appreciate yet, or that people might even disagree with?
Dedy: I think one area that would come to mind is if you look at, for example, the big labs now that are launching their products in this space, right? So you have the OpenAI Codex, you have Claude Code, right, from Anthropic. And I think the way they think about it is let the model do most, or keep the system layer very, very lean, and the model capability is getting better and better and better, and the model will solve everything eventually. Context windows are expanding, so you’ll just shove everything in a context window, and the model will do it.
We have a different point of view on that, a significantly different point of view. So we believe that for enterprise complex code bases, you’re not going to just shove the entire code base into the model context window for every inference of the model. You actually need to have a system that preprocesses the code base and creates their relationships and derives the insights. And you also want to give the developers the ability to control the agent, define the tools for different use cases, and create different workflows that are customized or configured for their specific use cases.
So we believe in a more controlled agentic environment where you have, again, I mentioned this earlier, a little bit a swarm of agent, and each agent is more tailored, and it has different permissions, it has maybe different tools, and this entire thing is controlled by developers.
And this is why we also believe that developers are not going away because they’re going to manage these agents, and they’re going to configure it and build them and track them and monitor them. So yeah, I think that’s probably the majority of the market think about this a bit differently, so is how we think about it.
Rolanda: Yeah, that’s a great insight. And thinking about outside of just Qodo, even maybe development lifecycle for a second, what’s a company or AI trend that you’re really excited about outside of all of this?
Dedy: Outside of coding, I’m very excited about the impact that AI can have in biology, for example, and potentially finding cures for diseases. I think the next couple of decades will be very, very exciting in this space. The big labs are going to scale reinforcement learning, that’s obviously in verifiable domains like coding. It is not even a question anymore. It’s obvious. And I think we’ll see in the rest of 2025 and 2026, a very significant rapid improvement in model capabilities because of the scaling of reinforcement learning that is going to happen. And I think if they’ll be able to solve this for other fields like biology and figure out how to close the reinforcement learning loop, then we’re going to see rapid advancements in these fields.
And I’m very excited about the possibility. There are still unknowns, a lot of unknowns there, but I’m hopeful. Obviously, it’s not my area of expertise, but I’m hopeful that they’re going to be able to figure this out and make significant advancements there.
Rolanda: I think that’s really powerful. Obviously, it’s great to impact people’s work, and that’s a lot of people’s lives, but obviously, there’s the actual life part of it as well. So I think that’s a great insight there. We’ve talked about advice that you would give others building in this space. What’s one piece of advice that you would give your own past self? If you were to rewind and think about when you were starting this company, what’s a piece of advice for what you would do differently?
Dedy: I think it’s a great question. I think to always remember that this is a marathon, not a sprint, and that’s in terms of the balance you need to strike as a founder. For example, for me, I used to be very big into rock climbing, and for the first two years of the company, I basically gave up on that because I couldn’t find the time … I couldn’t strike that balance. I started realizing that this is going to be a 10-20-year journey (who knows) doing this, so you’ve got to strike a balance. So, recently, I started getting back into climbing. And for me, it really affects me in a very, very positive way — and even makes me feel more productive at work. So it’s like you’ve got to strike that balance and realize that you can’t give up on things that are really important for you just because you’re a founder.
Rolanda: Yeah, I think that’s a really powerful message for people. And there’s, at least over here in the US, a lot of developers like to rock climb on the side too. So you never know. You might find some of your future customers there. So it can work out in both ways. And yeah, I mean, maybe just a fun question to wrap up with. You changed your name from Codium to Qodo. I’d love to learn what Qodo means.
Dedy: So Qodo is Quality of Development — and it’s like code with a Q. So the trigger to change from Codium, so there are. You’re probably aware there were two Codium companies. We both started at a similar time, and there was just a lot of confusion. Obviously, there’s an overlap. We’re more focused on quality, verification, and testing in enterprise organizations. So there was always differentiation, but there was still confusion. So that triggered the change. And I think it worked out quite well.
Rolanda: I love that Q for quality. I’ll remember that. Well, Dedy, thank you so much for the insights today, and thanks for joining us on the Founded & Funded podcast. I know I’ve learned a lot through our conversation, and I think it’s such a great story of your guys’ vision and journey, so I really appreciate you sharing that.
Dedy: Thanks, Rolanda. This was a lot of fun.