Fit for Purpose

Sign in Subscribe

tanner stahl

12,000 UFC Scorecards vs. Machine Learning And Human Judgement

How to evaluate a model when the ground truth is a panel of humans who disagree, and sometimes get it wrong. Twenty-five confident misses I trained a model to predict how professional UFC (Ultimate Fighting Championship) fight judges would score a fight from the round-by-round statistics. For

Evaluating A Viral AI Model: Build vs. Buy

A few weeks ago a new model called Jev launched, and within days it was being discussed across the industry, including inside our own organization. We spent the next few weeks evaluating it. What started as a question about whether to adopt Jev turned into something different. I think Jev

The Word "Hallucination" Is Doing Too Much Work

I spent the first stretch of my career in bioengineering, then in quality engineering on FDA-regulated medical devices. In that world, "biologically inspired" is not a finished thought, but more so the start of one. Before a biological analogy can justify anything about how a device works,

Zuckerberg's Infrastructure Argument

The claim On August 11, 2026, Mark Zuckerberg published a note arguing that the United States is losing the AI infrastructure race to China. His core evidence: "Countries like China are bringing online 1GW+ of nuclear capacity every other week, so we will need to accelerate building both energy

From 36% to 79%: Testing LLMs for Clinical Trial Matching

This is an engineering experiment, not a clinical validation study. ClinicalTrials.gov gives you age, sex, and condition as structured fields for every clinical trial, essentially for free. The eligibility criteria that decide whether a patient qualifies live in a free-text block: HbA1c thresholds, required medications, pregnancy exclusions, prior-

Architecting an Agent Around a Data Moat

The company below, Meridian, is fictitious. It isn't a real business, and nothing here describes an actual client engagement. Meridian is a B2B marketplace that connects buyers and suppliers across a fragmented industrial category. Its product is fine. Its moat isn't the product. It's

What Building Machine Learning Models Has Taught Me About Product Discovery

A cautionary tale for any PM about to greenlight an AI feature: sometimes the simplest approach works better than the complicated model. I work as a product owner for data science and AI products, where part of my job is figuring out which problems are worth solving with machine learning.

2026's Agent Sprawl Data

Ninety-seven percent of organizations say they're exploring agentic AI. Half of them rate their own capability as advanced or expert. Only twelve percent have a centralized platform to actually control what any of it is doing. That's from a 2026 OutSystems survey of 1,900

Neither of These Papers Cites the Other

Two pieces of research from 2025 started from very different ideas about how an agent can fail, and ended up pointing toward surprisingly similar practical controls. Here's the case for why they belong next to each other. The first is Anthropic's Agentic Misalignment research, which I

Your Agent Can Tell It's a Demo

The gap between an agent that works and an agent that's allowed to work is where most of the real engineering happens. Almost nobody talks about it publicly, because it's unglamorous and specific to whatever internal process a given company runs. But there's a