"The doctor said 'You're done with standard of care, there's no more treatment we have for you'... And then it was my 'oh' moment like oh, if I don't move I'm going to be dead soon."

In a deeply personal and highly technical conversation on the Changelog podcast, GitLab founder and former CEO Sid Sijbrandij shared the harrowing details of his battle with terminal bone cancer and his relentless pursuit of a cure. The discussion provided a profound look at how a master software engineer applies “first principles” to hack his own biology, while simultaneously launching a staggering 30 companies in four years—including his latest venture, the open-source AI coding platform Kilo.ai, and navigating the blistering economics of the AI compute revolution.

Hacking Mortality: A “Parallel” Approach to Cancer

In late 2022, just a year after leading GitLab to a highly successful IPO, Sijbrandij was diagnosed with osteosarcoma. A 6-centimeter tumor had grown on his spine, requiring immediate surgery to remove a vertebrae, followed by a spinal fusion with a titanium frame, radiation, and chemotherapy.

When the cancer returned in 2024, his doctors delivered a grim prognosis: he was out of standard-of-care options. Disqualified from several clinical trials due to a rare HLA type, Sijbrandij faced a critical turning point. “It was my ‘oh’ moment, like, oh, if I don’t move, I’m going to be dead soon,” he recalled. He stepped down as CEO to focus entirely on his survival.

Rejecting the “Single Variable” Medical Standard:

Sijbrandij approached his treatment the way a hacker approaches a failing system: by running diagnostics and testing patches in parallel. He flew to China for a rapid B7H3 scan and funded his own customized treatments. When a doctor warned him that combining untested therapies would ruin the scientific data—arguing, “what if it works, we won’t know what would have cured you”—Sijbrandij’s response was chillingly pragmatic.

“I let that doctor know that I wasn’t interested in finding what cured me, I was interested in getting cured,” he stated flatly. “I’m not research, I’m a real person.”

Single-Cell Sequencing and Radioactive Binders:

Through single-cell sequencing, Sijbrandij discovered that his tumor was heavily surrounded by fibroblasts (scar tissue). He leveraged an experimental treatment in Germany utilizing an FAP binder carrying a highly radioactive element (Lutetium/Actinium) straight to the tumor. The results were astounding: 60% necrosis and 20% shrinkage, which detached the tumor from his spinal cord and enabled surgical removal.

However, the treatment turned him into a walking hazard. “I was like 10 times as radioactive as an airplane at altitude from the inside,” he laughed. At airport security, he triggered massive alarms. “They had to call Washington because their sensors were showing plutonium.”

He also recounted a terrifying false positive where a PET scan showed 50 metastatic sites in his lungs. “I read my death sentence,” he said, recalling the moment he gathered his executive team to deliver the news, only to find out hours later that the spots were merely lingering COVID-19 scars. Today, his cancer is undetectable, though he remains vigilant with a “therapeutic ladder” of 30 backup medicines and an upcoming mRNA vaccine.

The Rise of Kilo.ai and the “Year of Parallel Agents”

Despite fighting for his life, Sijbrandij hasn’t slowed down. In March of last year, he launched Kilo.ai, an open-source, all-in-one competitor to tools like Cursor. He envisions a platform that handles the entire software lifecycle—coding, security reviews, and deployment—available across VS Code, JetBrains, CLI, and even a dedicated mobile app.

Sijbrandij highlighted the sheer velocity AI enables for modern developers. Recalling a recent “focus week,” he hired four engineers off LinkedIn on a Monday; by Friday, they had shipped initial prototypes.

“It used to be that you needed like a team of seven people to do something. Now you have one person and kind of a whole team of agents working on something,” he explained. “You ship at the speed of what two years ago would be a team of seven or eight.”

He predicts a massive shift in developer workflows is imminent. “I think 2026 will be the year of the parallel agents,” he declared. “It’s like the spinning plate act where you have all these agents up in the air… you continually have to kind of shake one of them to keep them all up in the air. You never have to be bored again or wait for them.”

The $100,000 Token Burn and the Multi-Model Future

A central theme of the discussion was the raw economics of agentic coding. Sijbrandij noted that while frontier models from OpenAI, Google, and Anthropic battle for dominance, a flood of highly capable, free models (like Grok, Minmax, and GLM) are emerging. Kilo.ai leans into this by supporting over 500 models, allowing users to orchestrate tasks across them without markups.

When challenged on the rising costs of AI tooling, Sijbrandij dismissed the idea that developers will be capped by a cheap monthly SaaS fee. He foresees a future of massive, individualized compute spend.

“The future we’re going to live in is not one where you have like a $20 subscription a month,” he predicted. “I think humans are going to burn $10,000 in tokens, maybe $100,000 in tokens per human… and they’ll be 100 times more effective than the humans of yesterday.”

AGI, the Ultimate Polymath, and the End of Collaboration

Reflecting on the pace of AI, Sijbrandij agreed with economist Tyler Cowen that Artificial General Intelligence (AGI) essentially arrived in April 2024, when models matched the median human intellect. For Sijbrandij, the true power of AI lies in its ability to act as the ultimate “polymath,” instantly bridging disparate fields.

He applied this directly to his cancer research. “We did an RNA test of my cancer so we know what RNA was going on, and we just sent the spreadsheet to ChatGPT, and what came out of it was really insightful,” he shared. Getting eight hyper-specialized doctors in a room is nearly impossible, but an LLM can integrate global medical knowledge instantly. “It’s as far as the eye can see every discipline you’re up to date on the latest knowledge.”

Closing the interview, Sijbrandij offered a piece of radical advice for modern knowledge workers: stop over-indexing on human teamwork.

“Because you now have a team of AIs reporting to you, you shouldn’t also be trying to partner up as much with other humans,” he advised. “Collaboration should never get in the way from making individual progress… just use your best judgment and get it over the finish line.”