Kevin Quennesson

When a patient has 28 lung nodules across 6 Chest CTs over 4 years (2022-2026), routine scan-to-scan text reports just say "stable."

We ran Gemini 4 Argon (8.74M tokens, 85 calls, 72 3D calipers) across all 6 scans at once:
braid.health/www/page/longi…

The agentic moment we saw in coding just happened in radiology.

Instead of scanning a file system, an agent can now scan the body in 3D:
1. look (up to 100 16-bit slices/batch)
2. annotate, analyze & verify across priors

braid.health/www/page/agent…

What happens when we run that same Left Shoulder MRI through Gemini 4 Argon with 6 million tokens of context?

26 interactive 3D callouts across all 8 MRI sequences, mapping both the tears and every intact structure in the joint:
braid.health/shoulder-argon

When you get a shoulder MRI, you usually receive 130 words of dense medical jargon ("low-grade interstitial partial tear of the infraspinatus") and zero images.

Your images, explained in plain English with interactive 3D MRI callouts:
braid.health/shoulder

Muse Spark is doing pretty well on medical images (crashlab.in/radle/2/leader…), so we're adding this soon on braid.health/ask-ai