Superposition Labs: the team building the deployment layer for medical AI.

Pablo Díaz grew up in Santiago watching his mother - a clinical psychologist - navigate a system where six months on a waiting list was optimistic. His family tree is dense with physicians and mental health professionals, the kind of people who treat healthcare dysfunction as a dinner-table topic rather than an abstraction. Before founding Superposition Labs he built software companies, but the gravitational pull was always medicine. (LinkedIn)

Víctor Perl comes from the same city and a parallel trajectory: a family of surgeons and doctors whose working conditions he observed long before he understood policy. He spent his career in engineering and operations roles where the core problem was always making complex systems reliable at scale - the exact constraint that now blocks medical AI from reaching patients. (LinkedIn)

We are co-founders with complementary obsessions: Pablo on product, regulatory architecture, and the clinical adoption problem; Víctor on infrastructure, reliability, and the operational surface that healthcare demands.

Chile has bimodal healthcare: world-class care for the privately insured, developing-world access for the 80% who depend on the public system. 2.6 million patients sit on the FONASA waiting list. The median wait for a specialist appointment or surgery is 242 days. Growing up in that bifurcation gives you a particular clarity about what broken access actually means - not as statistics but as people you know who cannot see a doctor.

The founding date - December 2025 - was not arbitrary. Medical AI crossed physician-level diagnostic benchmarks that year. Utah approved an AI copilot authorized to prescribe roughly 190 chronic medications. China deployed autonomous clinical AI across 260+ hospitals. ARPA-H funded the first agentic-AI clinical pilot in the United States. The regulatory walls are moving because the workforce gap is forcing them to move. We started the company at the intersection of those two curves.

We call it the harness: the deployment layer between frontier medical AI and the patient. Clinical integrations that connect to EHR systems and hospital workflows. Liability architecture that answers the question regulators ask first - who is accountable when the model is wrong? Data standards that let AI outputs plug into the formats clinicians and billing systems already speak. Regulatory scaffolding that did not exist eighteen months ago and has to exist within the decade.

None of this is the model itself. Frontier labs build the intelligence. We build the connective tissue that lets it operate in a clinical environment without breaking trust, compliance, or patient safety.

Google DeepMind, NVIDIA, Microsoft, OpenAI have secured immense funding, the world's best researchers, and a multi-year head start on foundation models. The gap in medical AI is everything around the models. No hospital can deploy a diagnostic system that has no liability framework, no integration with their EHR, no audit trail regulators will accept, and no workflow that clinicians will actually use. That surrounding infrastructure is unglamorous, jurisdiction-specific, and compliance-heavy - which is precisely why frontier labs are not building it themselves.

We wrote about this in detail in our founding essay and in The Deployment Gap. The short version: the bottleneck between a model that can diagnose and a patient who receives better care is not intelligence. It is deployment.

We climb in base camps. Each one is a product that solves a real problem for its own users and earns its own revenue, while quietly installing a piece of the harness in the ground behind it.

SignatureAPI is the first base camp: document infrastructure - electronic signatures, consent management, and compliance-grade audit trails. In healthcare context, it addresses an immediate operational pain point for clinics and health systems while building the data-integrity and identity layers that any future autonomous clinical AI deployment will require.

The radiology thesis is our next candidate. Diagnostic imaging is where AI performance is most mature, regulatory precedent is deepest, and the workforce shortage is acute enough that hospitals are already willing to adopt. Each base camp teaches us something about the deployment surface that no amount of theorizing could - and each one leaves behind infrastructure that compounds.

Write to us at contact@superposition.company. We are not hiring broadly. We are interested in clinicians who understand what breaks when you try to change a hospital workflow, regulators who have thought about AI accountability frameworks, and engineers who want to build post-superintelligence medicine rather than talk about it.

If you want to understand our thesis before reaching out, start with the founding essay, browse the FAQ, or visit the home page.