Founder Mode: On Cancer

In November 2022, Sid Sijbrandij — the Dutch engineer who co-founded GitLab and turned it into a $15 billion public company — was told he had osteosarcoma. A six-centimetre mass of bone cancer growing from his fifth thoracic vertebra. Rare. Aggressive. And in an adult, deeply unusual. Standard medicine had a playbook. Sid decided to write his own.

This is not a story about a man who got lucky, though luck played its part. It is a story about what happens when a systems thinker — someone whose entire career was built on running experiments in parallel, eliminating single points of failure, and ruthlessly iterating on data — turns that exact methodology on a disease that medicine typically treats with a sequential, cautious, one-thing-at-a-time approach.

And it is a story about what AI made possible that would have been impossible just five years earlier.

The Standard Playbook

Osteosarcoma in children and adolescents is rare. In otherwise healthy adults in their forties, it is vanishingly so. The diagnosis arrived without precedent in Sid's oncologist's experience — there was no well-worn protocol, no obvious referral pathway. Standard treatment was applied anyway: chemotherapy, surgical removal of the cancerous vertebra and replacement with a titanium fusion, stereotactic body radiotherapy, proton beam therapy, and an experimental Shasqi chemotherapy targeting technology that required an FDA Investigational New Drug approval to access.

It bought time. And then, in 2024, the cancer came back.

Facing recurrence, Sid did what founders do when the standard playbook runs out: he started writing a new one. He described the moment with characteristic directness in a post titled "I'm Going Founder Mode On My Cancer" — a deliberate echo of Paul Graham's essay on how founders operate differently from managers. Where managers iterate cautiously within known systems, founders act on incomplete information, run multiple hypotheses in parallel, and move faster than the situation seems to warrant.

"It became my own job to keep myself alive," he wrote.

Maximal Diagnostics

The first principle of Sid's approach was what he called maximal diagnostics: run every available test as frequently as possible. Not to decide anything immediately, but to build the richest possible picture of what his tumour was actually doing — not what osteosarcoma in general tends to do, but what this osteosarcoma, in this body, right now, was doing.

The diagnostic arsenal he assembled covered five pillars:

All of this generated an extraordinary volume of data. Sid uploaded it — all of it, including approximately 25 terabytes of genomic files — to Google Cloud buckets, then worked with AI systems to begin making sense of the patterns. Where an oncologist might review a single-cell sequencing report by reading the summary conclusions, Sid fed the raw data into AI tools alongside published literature on osteosarcoma biology, asking questions that no standard care pathway would ever prompt.

The FAP Revelation

The single-cell RNA sequencing changed everything. When Sid's team analysed the gene expression patterns across individual cells within his tumour, one signal kept appearing: elevated expression of the FAP gene — fibroblast activation protein. This protein is associated with the tumour microenvironment, the cellular ecosystem surrounding the cancer cells that either supports or inhibits immune attack.

FAP overexpression is not unique to osteosarcoma. But it is the target of a class of experimental radioligand therapies — drugs that pair a radioactive isotope with a molecule that binds specifically to FAP-expressing cells, delivering a localised radiation dose directly to the tumour. One such therapy, using the radioisotope Lutetium-177, was being administered at a clinic in Germany.

Standard oncology in Sid's home country would not have led to this option. It required reading the specific signal in his specific single-cell data, cross-referencing it with the published literature on FAP-targeted therapy, and making the decision to travel internationally for treatment that most oncologists had never considered for osteosarcoma.

In March 2025, following FAP radioligand therapy in Germany, imaging showed dramatic tumour shrinkage. The tumour was surgically removed in April. By June, scans detected nothing. In the space of eight months, a recurring osteosarcoma that had resisted aggressive first-line treatment had gone from detectable mass to undetectable — guided by a diagnostic insight that AI-assisted single-cell analysis had surfaced and that conventional care pathways had no mechanism to find.

The Timeline

What This Means for the Rest of Us

The story of Sid Sijbrandij is extraordinary — and it is also deeply uncomfortable, because it highlights a structural inequality in medicine that technology is rapidly making visible.

Sid had resources most patients will never have: the financial capital to access experimental testing and international therapies, the intellectual capital to understand what he was reading, the social capital to assemble a team of specialists around his case, and enough time — enough runway, as founders say — to run experiments in parallel rather than sequentially waiting to see if each one worked before trying the next.

But here is the thing about N=1 medicine done rigorously: it generates data that can help the next patient. Sid has published his full diagnostic data, treatment protocols, and outcomes at osteosarc.com. He has launched Even One Ventures specifically to try to scale this methodology — to build the infrastructure that lets the next person with an unusual cancer access the same diagnostic sophistication without needing a decade of experience running a tech company to navigate it.

This is the future of precision oncology that Sid's case points toward — not a future where exceptional individuals with exceptional resources get exceptional outcomes, but one where the methodology he pioneered becomes systematised, affordable, and accessible. Where the question is not "do you have the connections to access single-cell RNA sequencing?" but "what does your oncology platform show this week?"

There are serious AI-driven oncology programmes now running at major cancer centres that are beginning to operationalise exactly this approach. The convergence of CRISPR, mRNA vaccines, and AI analysis is creating a toolkit for personalised cancer therapy that simply did not exist five years ago. What Sid had to assemble manually, at enormous personal cost and effort, is being turned into clinical infrastructure.

He stepped down as GitLab CEO in December 2024 to become Executive Chair — a move that, whatever its operational reasons, also reflects the reality that keeping yourself alive had become, for a time, a full-time job. He now describes his mission in two parts: staying in remission, and making sure the approach that kept him alive is available to others.

He maintains monthly MRD blood tests. A personalised neoantigen mRNA vaccine is in development as an additional layer of protection. Engineered cell therapies with genetic logic gates — designed to engage only if the cancer returns — are being developed in parallel. The "Stay Paranoid" philosophy: never declare total victory, always have the next line ready.

Founder mode. On cancer. Still running.