Ostensible cofounders spent more time jockeying for board seats and credit from outsiders than mentoring a team, turning the place into a political snake pit that was later removed

Several things happened at Social Capital: A difficult divorce was underway during that period

Congratulations to the Perimeter Medical team and leadership.

From a systems perspective, this exemplifies real AI: invisible to the patient, indispensable to the clinician, and measured in fewer surgeries and a better treatment experience.

Initial focus will be on roughly 300,000 breast cancer surgeries per year in the U.S., with expansion into other solid tumors over time.

Claire is now a regulated tool that integrates into the surgical workflow in real time. This is just the start for what the platform can do for cancer care.

Claire’s real-time AI + OCT technology provides about 10x the resolution of standard X-rays, identifying suspicious tissue during surgery so surgeons can act immediately.

Approximately 20% of women currently face repeat surgeries because residual cancer isn’t detected until pathology reports arrive about 10 days after the operation, creating unnecessary waiting and anxiety.

Today Claire became the FIRST FDA-approved AI-enabled imaging device for breast cancer surgery and also received Breakthrough Device designation.

The goal was an AI-enabled device to help surgeons perform cancer operations more effectively: analyze a resected tumor in the operating room with AI and indicate immediately whether all cancer has been

About four years ago, roughly $40M was invested to help build Perimeter Medical, a small publicly listed company in Canada.

If those metrics are poor, a different delivery model may be needed

If teams still measure success by story points and sprint velocity, better metrics include defect rate, documentation coverage, time to onboard a new engineer, and incident frequency

Quality Is Speed Hours spent on rework, incident response, and technical debt are hours lost for upfront design and testing. Speed and quality need not be opposed — that tension is a false choice

Architecture is prioritized from day one and documentation is built into the process rather than added later

This approach, often called Software Factory, treats delivery like a production system with defined inputs, quality gates, and measurable outputs.

An Alternative: Software Factory Experienced engineers who focus on what actually works write specs, design systems, document decisions, and intentionally slow down to move faster later

The "Product Owner" role meant to represent customers often became a bureaucratic middle layer, distorting requirements at every handoff

With planning horizons of two weeks, engineering becomes feature assembly instead of system design. The result is fragmented architectures, microservice sprawl, and codebases no single engineer fully

The Standish Group's CHAOS Report found only 31% of software projects were deemed successful in 2020 What Often Goes Unnoticed Until It's Too Late Current development practices have eroded systems thinking