Insights

Thinking on AI systems,
clinical infrastructure, and compliance.

For teams building AI systems that need to hold up under patient data, compliance, and real-world use.

A high-fidelity digital transition from chaotic mesh to structured circuit architecture AI Architecture

From Vibe Code to Stable Code: Why Your Healthcare AI Prototype Needs a Production-Ready Core

Learn why moving from AI-generated vibe code to production-grade stable code is critical for healthcare AI companies and HIPAA compliance.

Read article
Sam Morhaim
6 min read
Cinematic illustration: a lone engineer standing at a junction where two paths diverge: a clean orderly path of structured blocks on one side and a dazzling tangled neon spiral on the other, against a dark navy background Strategy

The boring choice I didn't make

Years ago I picked the shiny database instead of the boring one that actually fit the problem. Here's what it cost me,…

Read article
Sam Morhaim
5 min read
Editorial illustration: a software engineer pushing a dumpster overflowing with papers toward a glowing teal AI engine AI Architecture

Congrats, your bad ideas now ship in record time.

Fable shipped a sprint in a day and a half that Opus 4.8 had been grinding through for weeks. The lesson is the old one.…

Read article
Sam Morhaim
5 min read
Editorial illustration: a row of teal-glowing test pipelines with one amber flagged mid-stream AI Architecture

How to Actually Test Whether Your Clinical AI Is Reliable

Most clinical AI teams test the wrong thing, the wrong way. Here's what reliability testing actually looks like —…

Read article
Sam Morhaim
11 min read
Editorial illustration: a small glowing teal gear at the center of a much larger machine sitting in shadow Strategy

The Model Was Never the Hard Part

Three studies dropped this spring. Three angles, one story. The model stopped being the hard part — everything around it…

Read article
Sam Morhaim
7 min read
Abstract visualization of a pristine source document being transformed into a fragmented, refracted output — representing the gap between correct retrieval and incorrect interpretation AI Architecture

Right Retrieval, Wrong Answer: The Failure Mode Most Clinical AI Teams Aren't Watching

Your RAG retrieved the right document. Your eval said retrieval was great. Then the model paraphrased a key qualifier…

Read article
Sam Morhaim
9 min read
Editorial illustration: a network diagram of nodes and data flows with one node highlighted amber Case Study

Two Weeks, One Map, and the Deal That Almost Didn't Happen

A founder with the biggest hospital deal of his life on the table and no way to answer the security questions. Two weeks…

Read article
Sam Morhaim
6 min read

Free 20-minute call

Building healthcare software?
Let's look at your actual tech.

Skip the corporate sales dance. On a free 20-minute call we'll dig into your biggest technical hurdle, map the path to fix it, and show you how to get to safe and scalable.