From GitLab to Kilo Code  ## Document Notes ## Highlights - **Rapid Response To A Life-Threatening Diagnosis** - Sid discovered a six centimeter tumor in his spine and underwent surgery, radiation, chemo, and an experimental single-patient IND treatment. - He traveled globally for diagnostics and combined treatments to scale options for others. Transcript: Sid Sijbrandij For sure. So end of 22, one year after the IPO, I discovered I had bone cancer, a six centimeter tumor grown from my spine. And we had to do surgery really quickly. I was in a lot of pain, and we removed the vertebrae and most of the cancer, and did a spinal fusion with a titanium frame. We did radiation, we did chemo, and I did my first single patient IND. Since 2017, I've invested in one other Y Combinator company. And over time, that guy became my best friend. His wife became my wife's best friend. And every time he was fundraising, the investors wouldn't put any money in. And every time I said, this doesn't make any sense to me that they're reluctant to put money in, I'll finance the company. So over time, I became his biggest investor and what he's doing is is combining drugs with binders through click chemistry allowing you to target different parts of the body and when I became so very sick he he made a we asked for an exception to treat me with his drug, and we got that exception. And we did all the treatments. And then two years later, in 24, the cancer started growing again. And the doctor said, you're done with standard of care. There's no more treatment we have for you. Maybe there's a trial somewhere. And I started looking around. I have a weird HLA type, which disqualified me from a few trials. And I have a bone cancer, osteosarcoma. It's kind of a rarer disease. So there wasn't anything for me. And then it was my moment. Like, oh, if I don't move, I'm going to be dead soon. I had to step down as a CEO because I needed to focus my full attention on this. I started talking to anyone, traveling anywhere. I went to the medical conferences. I went as far as China to get scans. And I started doing maximal diagnostics, making my own treatments, doing treatments in parallel, and trying to scale that for other people. And so that's been the journey I've been on over this last year. ([Time 0:07:04](https://share.snipd.com/snip/0f7ee333-81f6-47d8-912d-2c685abb0ff4)) - **First Principles Guide Personalized Combinations** - Combining treatments can be justified by first principles when large trials for combinations lack commercial incentive. - Multidisciplinary boards and histochemistry help safely combine therapies without standard trial evidence. Transcript: Sid Sijbrandij If if you want to be able to reproduce you need to isolate the chains so when i started doing parallel treatments one doctor told me to my face, but what if it works? We won't know what would have cured you. And then I let the doctor know that I wasn't interested in finding what cured me. I was interested in getting cured. That's right. I'm not research, but I'm a real person. That's right. The combination of treatments, there's very little incentive for pharma companies to kind of spend a billion dollars. It's now a billion dollars for a successful trial. To spend a billion dollars for an approved medicine to research a combination of drugs. So there's no incentive. So all you see on the market are these drugs taken in isolation. And doctors are reluctant to combine them because it's never been done before. But if you look at the first principles, you can say, hey, these treatments are probably okay to combine. What you don't want to do is have like combine two things that both hit the kidneys because your kidneys are going to be over capacity and that's not good so every time we do we combine treatments I have for example a pathologist in my tumor board to look at like hey can we combine these things and we also look at my own markup including genetics but also histochemistry where you Kind of we you color slides in and we do multiplexing so we can color multiple things and for example I told you I went to China China three months ago and I did that to get the first b7h3 scan Available to me and because of what we learned there we were able to kind of change the development of one of the drugs I'm developing to remedy what we saw. We saw in China that I have higher B7H3 expression in the liver than any one of the 20 patients they saw before. And we decided to combine that medicine with something that's rarely expressed in the liver. So it's not going to destroy my liver going forward. So you got to be kind of, what I'm trying to do is kind of be first principles about it instead of kind of relying on evidence for the ability to combine because that evidence, it doesn't Exist because no one has the incentive to do that. ([Time 0:10:36](https://share.snipd.com/snip/75bfcc57-100b-47b0-b518-3b6a29298f91)) - **Therapeutic Ladder Buys Time** - Sid currently has no detectable cancer but assumes microscopic disease remains and maintains a therapeutic ladder of about 30 medicines. - He develops multiple drugs and diagnostics simultaneously to keep treatment options open. Transcript: Sid Sijbrandij Right now we cannot detect my cancer. So that's great. It's a really good spot. That is great. Yeah, that's a great spot to be. Compared to a year ago where it was growing and we had zero treatments. Today I have kind of a therapeutic ladder as in like if things get worse, we'll escalate the medicine. And there's 30 medicines on it. So like we have options now. ([Time 0:14:29](https://share.snipd.com/snip/4a35ac29-8870-40db-81e2-9e8d6315c116)) - **Targeted Radioligand Enabled Curative Surgery** - Single-target radioligand therapy bound to FAP scar tissue caused 60% necrosis and 20% shrinkage after two treatments. - The shrinkage allowed surgeons to remove tumor tissue previously impossible to reach safely. Transcript: Sid Sijbrandij And then on the other side, you put a radioactive element, lutetium or actinium. And it did two treatments there with incredibly good results. First of all, no side effects. I couldn't detect any. And we saw 60% necrosis. And 20% shrinkage, which considering it was only two treatments, that was a great result. And my cancer is in my back near my spinal cord. Because of the shrinkage, it detached from the spinal cord, allowing a really talented surgeon to go in and go after it and remove most of it surgically, which wasn't possible before. ([Time 0:16:35](https://share.snipd.com/snip/6406845e-65e8-494f-b602-79141bffadae)) - **Preserve Samples By Flash-Freezing Tissue** - Ask hospitals to flash-freeze surgical tumor samples to preserve them for later diagnostics. - Flash-freezing retains options for advanced tests that FFPE would preclude. Transcript: Sid Sijbrandij For people with cancer, it's way better to flash freeze stuff because it allows you to to do many, many more diagnosis on it later than if you do the standard, which is FFPE, which preserves It, but it precludes you from doing a ton of tests. So if you want to have an active role in your treatment, you've got to convince the hospital to flash-free some of it so that you preserve optionality to run these diagnostics later. And what we saw was that the T cell infiltration went up a lot. It went up from 20% to 90%. So the place was buzzing with T cells, which is a good thing. And we think that is a combination of the two checkpoint inhibitors that I'm taking that kind of unleash your immune system. And one experimental treatment that is an oncolytic virus, so like modified common cold virus that drops the TGF beta. And the TGF beta is something the cancer uses to hide from the immune system. And it kind of tears away that invisibility cloak. So that's good news. We don't super think that that's curative. But I'm gearing up in two weeks to take an mRNA vaccine. And that's like a 50-50 of curative. But there's a reason we have like 10 drugs in the pipeline. ([Time 0:27:35](https://share.snipd.com/snip/9a1c7928-0a89-4d80-ae92-f1f0e72381ac)) - **Ship Fast By Empowering Individuals** - Focus on building products that let individuals move fast and ship end-to-end rather than endless coordination. - Use AI agents and contributor communities to accelerate feature development. Transcript: Sid Sijbrandij Yeah, it's a lot of working. I like working. These last few weeks I've worked 10-hour days. A lot of meetings, mostly the whole day is meetings, 25 minute meetings, like 15 a day. I've also done some traveling for pleasure. We thought at the beginning that this might be my last year. So we celebrated the 25 year anniversary of me and my wife extensively. Thank you. With friends and family. That was great. Also a lot of traveling, like going to medical conferences, going to treatments on multiple continents. But also doing fun stuff, like adding observability to GitLab. I started an AI accounting company called Kilo. I'm still creating these companies around open source projects with OpenCore Ventures. And then since September, starting two biotechs every month. ([Time 0:32:42](https://share.snipd.com/snip/22ddb4a9-82fb-4dbc-8fc2-3fbf9163b7de)) - **All-In-One Agentic Engineering Platform** - Kilo aims to be an open-source, all-in-one agentic engineering platform covering coding, review, security, and deployment. - The open-core model charges managers for org-level features while keeping core agentic tooling open. Transcript: Sid Sijbrandij And obviously agentic coding is very cool. It's the future. And it was cool to see that open source had a chance here. I kind of, before I thought it's all going to be closed source and I saw the rise of open source and I thought, okay, time to join that game. Adam Stacoviak Are you deploying a similar playbook that you deployed with GitLab back in the day? Yeah. Sid Sijbrandij Kilo has great agentic coding, but the best reason to use it is that it's all in one. So not only can you do the coding, it can also review your code. Not only can it do that, it can also do the security review. Not only can it do that, it can also deploy it. So it's available everywhere. VSCO, JetBrains, CLI, in the cloud, in the mobile app. And it can do the full lifecycle, including deployment, and we're working on more there. So just like GitLab, the idea is to be the all-in solution where you don't have to switch applications the whole time to do what you need to do and to be open source where people can contribute Improvements. Adam Stacoviak Right. So what's the open core aspect with Kilo? ([Time 0:34:11](https://share.snipd.com/snip/141525d7-e014-42c3-9383-f9ad8bc575b5)) - **Multi-Model Strategy Over Single-Provider Lock-In** - Kilo stays model-agnostic, supporting hundreds of models including free frontier models as they emerge. - Users pick models per task and share profiles for consistent, team-wide agent behavior. Transcript: Sid Sijbrandij We want to, we're not working on models ourselves. We're working very, very hard on supporting all the models. So hundreds of models work over 500. We're the most prominent launch partner for new models. So a lot of new models, a lot of stealth models are first launched on Kilo. Adam Stacoviak And our users, a lot of our users use a models like grok min max glm uh with uh with kilo interesting i think i haven't even heard of a couple of those i've heard of grok but the min max i haven't Even heard of so this is either underground or upcoming or when you say new bleeding edge perhaps are these oh yeah it's all new it's like all like launched in the last two months there's Sid Sijbrandij A lot of kind of chinese frontier labs and they they want to quickly get their models more popular so they offer incredible pricing just like like grok it's it's it's got code is incredibly Jerod Santo Good model and it's been free for months now it's an interesting space because you've got a lot of things happening around google in particular there was an announcement today for a Flash model i think it's called gemini flash if i can recall correctly but uh you know when you look at that they can essentially offer their models in the API layer for virtually free Or a loss leader because they're trying to build their platform. While OpenAI and Anthropic are burning cash, making money for sure. But, you know, when you're fighting a competitor, that essentially can outlast you because they already have so much incumbent cash or value elsewhere in the chain. It's kind of hard to battle somebody who's willing to lose or break even when you have to profit. What's your take on the current landscape of just the competitive nature from nation states like maybe China, etc. Sid Sijbrandij Maybe that's just companies there to the battle that's playing out here in the US. Yeah, I think the kind of the leading coding model, that's an incredibly competitive space where kind of it always feels a bit like OpenAI has a better model. And then Google has a better model and then Entropic comes back with like the best model by far, by for sure. Yeah. But also the, the best kind of free model is incredibly competitive with deep seek, min, max, DLM, all these providers, uh, um, kind of one upping each other. And what we offer our users is you're going to keep using the same application. You don't have to sign up for a ton of stuff. It's just our thing. There's no charge. It's completely free. And then you have the ability to use any model you want. Soon you can use different models. You can use free models, bait models, whatever you see fit. ([Time 0:36:16](https://share.snipd.com/snip/796b51fe-dba8-4d0c-a2eb-7a577de04269)) - **Parallel Agents Multiply Developer Output** - Parallel agents plus small context windows let you split big tasks into many cheap, scalably run jobs. - Running multiple agents and model runoffs increases robustness and speeds delivery. Transcript: Sid Sijbrandij So, yeah, we have like 20 engineers. But in reality, they're all working with seven agent decoders. And that's the expectation that you, as one human, you ship at the speed of what two years ago would be a team of seven or eight. Adam Stacoviak That's interesting. So you've successfully parallelized your cancer treatments and your engineers are parallelizing their agents. I can still only keep one agent busy at a time. Maybe I'm not ready for this new era. I couldn't work for Kilo. How do they do it? You know, when you got all these different, you know, down in the weeds of how they're managing multiple agents, working on a singular feature, I suppose, or a set of features that support A larger feature, and how are they staying productive and fast in that environment? Sid Sijbrandij That's common. Like you're not alone in only being able to keep one agent busy. But I think 2026 is going to be the year that that changes. Our parallel functionality where you have kind of, you farm it out to multiple agents in multiple work streams. That's just getting started. A month ago, the engineer who made Schaltwerk, one of the early ways to parallelize things, joined us and he's doing amazing work. There's still a lot to do. For example, Kilo has a functionality that's orchestrate. With orchestrate, you say, this is what I want to achieve. And split it up into separate jobs. So each of those jobs has small context windows. Because small context windows means it's cheaper to run it's faster to run but most importantly it's not going to lose lose half of its context because it's it's kind of getting to to The the maximum context window that's effective but in the coming weeks we'll ship something where those different agents are now able to work in parallel. And you can do a runoff, especially if you're using these three models. You might as well have two models work and then see which one has a better result. So I think we're going from you have one working directory today to in the future, you have multiple tasks running in parallel. ([Time 0:42:26](https://share.snipd.com/snip/57b9c49f-9c7b-4d1a-b46b-5150cad5c3e4)) - **Use Free Models For Iteration, Paid For Depth** - Favor free or low-cost frontier models for iterative tasks and save expensive models for deep reasoning. - Combine free models with paid models at list price to control token costs transparently. Transcript: Sid Sijbrandij There is no catch. Adam Stacoviak Grok is amazing. I've been told that plenty of times in my life, and then eventually there's a catch somewhere, somehow. Sid Sijbrandij Well, it might be that this whole thing ends at some point, where these models... Adam Stacoviak Yeah, the catch is it's temporary maybe. Sid Sijbrandij Where these models. But what we've seen over time, this has been going on for half a year now. And there's models that kind of launch for free. And at some point they become paid. But the rate at which free models are launching is faster than the rate at which they're getting deprecated. So there's now more free models on the market than there were half a year ago. ([Time 0:45:18](https://share.snipd.com/snip/a794299f-faed-4ee9-8a40-6c92445e930a)) - **AI As The Ultimate Polymath** - Sid argues AGI-level capability arrived for median tasks around April, enabling AI as a broad polymathic assistant. - That rapid capability shift makes AI valuable for integrating knowledge across many specialist domains. Transcript: Sid Sijbrandij And it happened in April. Like it just flew by and we hardly even noticed it. You said we had AGI in April. Adam Stacoviak Is that what you just said? Yeah. Sid Sijbrandij I think in April, the models became as smart as the median human. Oh, is that what AGI is? Artificial general intelligence, yes. My definition, I stole April from Tyler Cowen, which I think is one of the smartest people on Earth. Okay. Adam Stacoviak I must have missed it. I just don't know how generally, I guess maybe you're saying because most humans are generally unintelligent. Sid Sijbrandij Well, also like think about chat GPT, like it's, it, it might not know more than the expert in a field, but it knows it is, it knows a lot about everything. The ultimate polymath. Jerod Santo Yeah. Yeah. I mean, like I was having some really interesting conversations with chat GPT the other day that was, that was very therapeutic. ([Time 0:49:18](https://share.snipd.com/snip/d492f638-5072-4cc0-ba48-06e64075529a)) - **Use LLMs To Accelerate Medical Reasoning Safely** - Use LLMs to integrate literature and generate diagnostic hypotheses, but always verify with clinicians before treating. - Treat models as powerful thought partners, not sole decision-makers for care. Transcript: Jerod Santo Can you be specific in any way? Sid Sijbrandij Like what, what blew your mind, I suppose? What you're, what you're looking for with RNA is like not so much that you have the genetics, but what's actually happening at the cell level, what's being expressed and what are potential Pathways that are overexpressed and you can target. Because I know from the genetic, there's something that sounds like a party drug, but isn't. MDM2 is overexpressed. But I want to see how the thing it acts on, the P53 division, how that's going in my cell. And there's better examples here, but I'd have to pass the GPT for them. I don't want to be distracted while talking with you. But like, for sure, just incredible things. And it's not the end all be all. Like, don't take a treatment without first checking it with a doctor. ([Time 0:51:53](https://share.snipd.com/snip/682ab465-5191-45d1-aac8-a6a2aae94e08)) - **Hire Doers Who Ship Independently** - Hire people who can independently get things done and ship without excessive coordination. - Encourage individual progress and use AI agents to reduce human collaboration overhead when possible. Transcript: Sid Sijbrandij So OpenCore Ventures has a portfolio on their website. And I think companies always need people who can get things done themselves and are able to go after something by themselves and get it done, get it over the finish line. At one of the post hoc founders wrote a great article about stop collaboration. Like there's, there's a lot of, because you now have a team of AIs reporting to you, you shouldn't also be kind of trying to partner up as much with other humans. Like you have all the ways to make it happen. In order to go fast, collaboration should never get in the way from making individual progress. It should never delay kind of trying to get something over the finish line yourself. ([Time 1:13:18](https://share.snipd.com/snip/73ef6b0c-bdf4-4ea2-9f44-5eb507883a58))