Blog

AI labs are surrounding their models with proprietary tooling — subscriptions, IDEs, agents. The bet is that you'll never leave. Here's why staying model-agnostic is the safer play.

I threw a raw astrophotography FITS file at VerbaGPT and asked it to process the Orion nebula. It applied color stretching and star-masking in 15 minutes. Code + LLM is unreasonably effective.

AI makes it trivial to build software for narrow problems. The result is a tool zoo that demands more attention than it saves. The solution is one-to-many problem solving.

A walkthrough of what fully autonomous data analysis looks like when it works — and where it still falls short.

We spend a lot of time asking what AI can do. Not enough time asking what we actually want it to do.

Testing LLM agent skills on a weather analysis problem — what worked, what surprised me, and what still needs work.

LLMs are surprisingly good at analysis — until the data is messy. Here's what I found testing against real-world healthcare data.

Offloading thinking to AI can make you faster. It can also make you shallower. The tradeoff is worth understanding.

There's a school of thought that scaling compute is all you need. Here's what that gets right — and wrong.

Small percentage fees have enormous compounding effects over decades. An interactive calculator to see exactly how much.

Actuaries are trained to quantify uncertainty. But there's a difference between modeling it and truly embracing it.

Complex risk adjustment models don't outperform simple ones at the plan level. The data is clear — and the implications are significant.

Benford's Law says the first digit of real-world numbers follows a predictable distribution. Healthcare billing data does not always agree.

AutoML promises to democratize data science. A look at where it actually delivers and where you still need a human in the loop.

The promise of AI in healthcare is enormous. The reality — in 2020 — is more complicated. A look at where we actually were.

Social determinants of health, genomics, and behavioral data — what they could mean for risk adjustment models.

Risk adjustment models look backward to predict forward. Here's what that tension means for payment accuracy.

A walkthrough of ACA RADV — what it is, how audits work, and what health plans need to know.

AI has gone through multiple winters and springs. What's different this time — and what's the same.

The ACA risk adjustment program hit pause in 2018. Here's what happened and what it meant for the individual market.