field notes

Signals from the perimeter.

Social posts scroll away. This is where the thinking lives: permanent, indexed, and ours. Notes on memory, detection, and the worlds inside the nugalaxy universe.

latest

A holographic AI agent enclosed in a diagnostic measurement rig, probe lines reading its behavior, one probe glowing warning-lime where a reading has drifted

Jul 20, 2026ai-agents

A Free Guide to AI Agent Evaluation

340 tests passed on Friday. Monday brought 43 tickets. I wrote a free short book about AI agent evaluation, and why your test suite cannot see this coming.

The holographic agent running steadily on a laptop-sized slab. Beside the slab, an oversized, dark, and idle GPU tower sits in the void

Jul 17, 2026performance

You probably don't need a GPU for that workload

Reaching for a GPU is usually sizing for the workload you imagine, not the one you have. A test for whether you need a GPU, and why the answer is often your laptop.

Agent focusing on mismatched data shapes

Jul 16, 2026staging

Staging vs production: the gap was always the data

Staging matches production on schema and infra, then lies about the data. Why the staging vs production gap is about shape, not size, and how to close it without copying prod.

It depicts the agent interacting with a floating digital interface, attempting to write while surrounded by the complex HUD

Jul 15, 2026writing

AI writing tells: the real one isn't the em-dash

Everyone hunts for em-dashes and the word "delve." The durable AI writing tell isn't vocabulary, it's the absence of specificity and stakes, and why that still matters.

Agent standing steady on its own isolated data block while neighboring cubes wobble on a shared shifting slab, on the void

Jul 14, 2026flaky-tests

Flaky tests are a data problem, not a timing problem

Retries and sleeps hide flaky tests, they don't fix them. Why most flaky tests are a data problem, not a timing problem, and what actually stops the flake.

It shows our agent holding that single, elegant cube against the sprawling, messy scaffold

Jul 13, 2026over-engineering

Over-engineering is a bet on a future you can't see

The simplest thing that works looks junior, so smart engineers skip it. Why over-engineering bets on a future you can't see, and simple stays cheap to change.

all posts

A holographic AI agent enclosed in a diagnostic measurement rig, probe lines reading its behavior, one probe glowing warning-lime where a reading has drifted

Jul 20, 2026ai-agents

A Free Guide to AI Agent Evaluation

340 tests passed on Friday. Monday brought 43 tickets. I wrote a free short book about AI agent evaluation, and why your test suite cannot see this coming.

The holographic agent running steadily on a laptop-sized slab. Beside the slab, an oversized, dark, and idle GPU tower sits in the void

Jul 17, 2026performance

You probably don't need a GPU for that workload

Reaching for a GPU is usually sizing for the workload you imagine, not the one you have. A test for whether you need a GPU, and why the answer is often your laptop.

Agent focusing on mismatched data shapes

Jul 16, 2026staging

Staging vs production: the gap was always the data

Staging matches production on schema and infra, then lies about the data. Why the staging vs production gap is about shape, not size, and how to close it without copying prod.

It depicts the agent interacting with a floating digital interface, attempting to write while surrounded by the complex HUD

Jul 15, 2026writing

AI writing tells: the real one isn't the em-dash

Everyone hunts for em-dashes and the word "delve." The durable AI writing tell isn't vocabulary, it's the absence of specificity and stakes, and why that still matters.

Agent standing steady on its own isolated data block while neighboring cubes wobble on a shared shifting slab, on the void

Jul 14, 2026flaky-tests

Flaky tests are a data problem, not a timing problem

Retries and sleeps hide flaky tests, they don't fix them. Why most flaky tests are a data problem, not a timing problem, and what actually stops the flake.

It shows our agent holding that single, elegant cube against the sprawling, messy scaffold

Jul 13, 2026over-engineering

Over-engineering is a bet on a future you can't see

The simplest thing that works looks junior, so smart engineers skip it. Why over-engineering bets on a future you can't see, and simple stays cheap to change.

The Techwear Architect A character in sleek cyberpunk gear or techwear, calmly placing a glowing block onto the rising staircase

Jul 12, 2026developer-audience

Trust compounds, attention doesn't: build a developer audience

A viral post is a lump sum that decays. Trust is an interest rate. Why your developer audience is a rate you compound, not a number you chase.

high-tech agent scene holding microphone

Jul 11, 2026building-in-public

Building in public is backwards: build first, then speak

I used to post the roadmap and the coming-soon teaser. Then I stopped. Why building in public is backwards, and the rule that beats it: build first, then speak.

a lone robotic hand and HUD elements, focusing the hand on the central void

Jul 7, 2026design

Design the empty state first, not the happy path

Eight years as a product designer taught me one habit that changed how I write code: design the empty state first. Why the states you skip are the product.

Agent inside holographic shield, representing an agent's bounded blast radius

Jul 5, 2026ai-agents

AI Agent Guardrails: Contain the Failure You Can't Prevent

You can't stop an agent from ever being wrong. AI agent guardrails, human-in-the-loop, and blast radius: how to contain the failure you can't prevent.

AI Agent Cost, featuring agent with holographic coin

Jul 4, 2026ai-agents

AI Agent Cost: The Budget That Decides If It Ships

An accurate agent still dies in review if it is too slow or too expensive. How to budget AI agent cost and latency per successful task, not per call.

It features the agent analyzing a holographic timeline of unique event shards

Jul 3, 2026ai-agents

AI Agent Reliability: Measuring What Won't Repeat

An agent won't give the same answer twice, so one green eval run is a coin flip. How to measure AI agent reliability as a distribution, not a pass.

It portrays the agent interacting with a structured library of holographic data shards and cubes

Jul 2, 2026ai-agents

AI Agent Evals: Your Eval Set Is the Product

Most teams treat AI agent evals as an afterthought. Your eval set is the product: how to build failure cases that survive every model swap.

Telemetry readouts, scanning reticles, and data streams to emphasize the validation theme for agents

Jul 2, 2026ai-agents

Is Your AI Agent Production-Ready? Define the Bar First

"Is it ready?" is a circular meeting until you define the bar. The four lines that make a production-ready AI agent a number, not a vibe.

An empty wireframe grid filling with ordered lime data blocks emitted by a single electric-lime cube.

Jun 30, 2026python

The empty-database problem: realistic test data in one command

The expensive part of test data is not the rows. It is deciding what a good row looks like. How to seed a database with realistic, reproducible synthetic test data in one command.