Building Is No Longer the Bottleneck
AI has made software dramatically cheaper to build, but not customers, validation, or attention easier to earn. When another feature is the comfortable option, the hard work begins outside the editor.
Read more →Blog
Thoughts on AI agents, automation, and building systems that work
It Was Never the Restriction
I took away an agent's ability to run commands, then its access to the registry, and measured what that costs. Then I ran the same matrix on two weaker models. The restriction cost 26% more tokens; the weakest model never once went to look, and filed fourteen reports claiming a release that didn't exist.
Read more →The Instrument Fails in Your Favour
Across three studies, every measuring instrument I built broke at least once — and every break pointed toward the result I was expecting. On why 'just measure it' is incomplete advice, and what building the reference actually costs.
Read more →Stop Being the Cable
Most teams that try an agent and quietly go back weren't short on skill. They were doing the plumbing by hand — copy, paste, carry the answer back — and that costs the same every single time. What connecting it actually involves, and which part of it you'd have to build yourself.
Read more →An LLM Can Infer the Rule You Forgot — in One Dimension
My last preprint concluded that LLMs translate rules and don't infer them. In continuous control that turns out to be false for a wall and true for a circle: the model repairs an omitted 1D rule in 105 of 111 attempts, and never once recovers the 2D version — through eight interventions designed to fix it.
Read more →Nobody Will Check Behind You
Eight lines in a brief took a coding agent's detection of its own false release from 8/15 to 15/15. The clause never says where to look — only who answers for what got published.
Read more →The Tool You're Allowed to Use
Every piece of advice about working with agents assumes you picked the tool. Plenty of people didn't. Here's what changes when the choice isn't yours — and the argument to make if it partly is.
Read more →The grader knew less than the system it graded
An external report said the classifier was 54% accurate over 500 companies. The gold set had been generated by a general-purpose model with all 500 at once, and its labels could be guessed from the company name alone. The defensible number was twenty points higher. On when an AI can evaluate another AI, and when it can't.
Read more →What Coding Agents Say When They Talk to Each Other
I read 179 real messages between parallel Claude Code sessions: the channel is almost never used to ask for things, but to tell the other one something true about their own work. Then I built the experiment — and watched one session catch another shipping a release that did not exist.
Read more →If You Came From Engineering
You already know how to verify work at scale — tests, CI, pipelines. You're just not applying any of it to the agent. You're checking its output with your eyes, which is the one thing you spent a career learning doesn't work.
Read more →The Scaffolding You Pay For
I was convinced that prescriptive skills get in the way of a frontier model, so I measured it: ~640 responses and 147 agent runs. I was wrong about where the damage is — and once the agent has tools, the benefit disappears and only the bill stays.
Read more →One Grey Level Out of 255
I rendered text into images and faded it until I couldn't read it, then asked six vision models and two OCR engines to try. They read text at a contrast of one grey level out of 255 — invisible to me — and some of them obeyed instructions hidden there. The margin they have over me turns out not to be sharper eyes.
Read more →Your Agent Doesn't Know What Has Already Happened
Agents plan around events that never occurred and invent dependencies between things that don't depend on each other. It looks like a missing clock, so I measured it: 572 responses, six models, two domains. The date changes nothing, and one failure gets worse in the strongest models.
Read more →If You're Starting From Zero
You described what you wanted, an agent built it, and it works. Here's what to learn first — not the whole list, the order. Written for people who never wrote software and don't intend to start.
Read more →What You Still Need to Know to Ship
You can build software now without writing a line of it. What's left isn't coding — it's a way of working, and a list of things you have to know exist or you'll never think to ask. Here's both.
Read more →Confident About Text It Cannot Read
A safety filter gave a base64 string a medium self-harm severity. Two judges scoring the same ciphered replies agreed worse than chance. A base model wrote fluent cipher it had never understood. Three systems producing confident verdicts about text they cannot read — and one conclusion of mine that didn't survive the full run.
Read more →Where's the Ball? Part 4 — When Nothing Moves
A coda to the trilogy: chasing the specialist's blind spot led somewhere I didn't expect. The hardest ball to locate isn't the one flying through the air — it's the one sitting perfectly still.
Read more →Three Judges, Three Rankings
I ran the same 45 blinded comparisons past a GPT, a Grok and a Claude judge. Each produced a different ranking, each favoured its own answers, and the effect scales with how subjective the task is.
Read more →Where's the Ball? Part 3 — It Was Geometry All Along (and the Model Doesn't Know When It's Wrong)
Opening up the tiny network that beat the frontier models: ten interpretable geometric numbers recover 92% of it, fancy topology adds nothing, and the model turns out to be overconfident on exactly the cases it gets wrong.
Read more →I Built an Expensive Form
I argued that a conversational journey added little over a well-designed form. The call went the other way, so I built it — and the client's feedback settled the argument better than I could have.
Read more →Where's the Ball? Part 2 — a 60-Kilobyte Model, Two Sports, and the Signal VLMs Miss
A tiny network trained on my laptop solves the hidden-ball cases that frontier VLMs can't — and a two-sport transfer study reveals what, exactly, a model learns about a game.
Read more →Routing Engineering
Once prompting stops being the hard part, a new question takes its place: which model, at what reasoning effort, for which slice of the work. Choosing well is becoming its own engineering discipline.
Read more →The Death of the Prompt Engineer
A claimed counterexample to a 30-year-old conjecture emerged from prompts like 'do a breakthrough' and 'just continue.' The craft of writing the perfect prompt is dying — and something more interesting is taking its place.
Read more →Where's the Ball? Testing Whether a VLM Has a Spectator's Intuition
I hid the ball from vision-language models and asked them to find it from the players alone. The answer was interesting — but the more interesting part was how easily I almost concluded the wrong thing.
Read more →Bring Your App to the Agent
The next paradigm shift in application development isn't putting an agent inside your app — it's putting your app inside the agent your users already have. MCP is just the current plumbing for it.
Read more →Your Agent Doesn't Know What's Internal: Context Leakage in AI Workflows
AI agents keep quoting internal feedback, spec codenames, and process notes in client-facing deliverables. Why it happens, a taxonomy of the three failure modes, and the structural fixes that actually work.
Read more →Writing a Research Paper with AI: Where Model Intelligence Actually Shows Up
The real process behind an AI-assisted preprint — and why the hardest frontier isn't prose but planning the science, where a more capable model stops being a convenience and becomes the thing that lets the work go further.
Read more →Teaching a Model a Language You Just Invented
Speak to a language model in a cipher it has never been told the key to, and something surprising happens before any decoding does: some models refuse to play at all. A small experiment on how fast models crack novel codes — and where that runs into a safety wall.
Read more →A World Model Can Pass Every Test and Still Lose
I set out to reproduce a DeepMind result and instead found a clean way verification can lie to you: a code world model that passes its gate at 100% accuracy, stays 98% accurate on the states a planner visits, and still loses systematically at play.
Read more →Let the Loop Iterate: Restarting Only Pays If You Know What to Throw Away
Karpathy suggested letting a stuck agent restart instead of grinding on. I built a loop with visible tests and measured it: iterating on feedback beats clean restarts at equal compute — and handing the agent a scoped restart button reveals that knowing *what* to discard is what separates the tiers.
Read more →Repetition at the Edges of Language
Say the same thing to a language model far past what any human would tolerate, and the polished ones don't break — they quietly give up. Strip the alignment layer off, though, and the raw model degenerates into loops. A small experiment on what repetition reveals.
Read more →Coding Agents and Teamwork: Social Skills, or Structure?
A recent benchmark says coding agents can't collaborate because they lack social intelligence. I test a different hypothesis: they lack structure — and find that making integration one agent's job recovers collaboration, while merge conflicts mostly mask deeper semantic failures.
Read more →Bootstrap the Environment, Not the Agent
Why cloud workspaces and AI coding sessions need repository-owned bootstrap scripts for system packages, tests, and builds.
Read more →The Forgetting You Don't Measure
When you continually pre-train an LLM into a 'world model', it quietly forgets general knowledge — and a little data mixing buys most of it back. But how much it forgets depends heavily on how you specialize it, and full fine-tuning pays three costs a single benchmark undersells.
Read more →The Only Software You Need Is a Subscription
Agents made building almost free, which exposed what you were actually buying all along — not the software, but the maintenance, the de-risking, and your attention. A build-vs-buy calculus rewritten around the one resource that doesn't parallelize.
Read more →Results-Oriented Programming
Reviewing used to mean reading the implementation. Increasingly I verify what the model produced, not the path it took — and what 'verify the result' means changes wildly depending on whether the result is a number or a behavior.
Read more →How Much Should You Still Know?
I build systems whose business logic I couldn't recite if you woke me up at 3am. The model holds it for me — and that's fine. The hard question isn't how much you delegate, it's whether you can still tell when the answer is wrong.
Read more →Demos That Wow vs Demos That Work
Two philosophies of the client demo — mock-first spectacle vs functional proof-of-concept — and why, with AI, the prettier one hides exactly what matters: latency and non-determinism.
Read more →When the Chat Builds Its Own Interface
Generative UI lets an agent render interactive widgets — forms, charts, signatures — inside the chat instead of plain text. AG-UI is the protocol standardising it, and I've already been building it by hand.
Read more →LLM-as-Judge Is Three Decisions
Before you write a judge prompt, you owe three decisions: the context, the unit, and the dimension. Here is the framework, applied to evaluating my production DevOps agent.
Read more →Building a Production DevOps Agent: From Slack to Kubernetes
How I built an AI agent that investigates infrastructure issues autonomously — with read-only guardrails, output compaction, and infrastructure knowledge baked into the system prompt.
Read more →From Claude Artifact to Production PWA: Building VitaminD Explorer
How a simple question about vitamin D synthesis turned into a Claude artifact, then a full PWA with D3.js visualizations, push notifications, and 6 languages.
Read more →Software Is Dissolving Into the Model
What we write is becoming markdown skills, what users see is becoming generated pixels, and the middle keeps getting thinner. A look at where software is heading in 2026.
Read more →The Real Skill in the Age of AI: Knowing When to Stop
AI can now build almost anything. The hard part is no longer the code — it's managing your own cognitive limits as you orchestrate multiple agents in parallel.
Read more →Playwright CLI vs Scripts: How AI Agents Should Test
Why most AI-driven tests should be reproducible scripts, not CLI exploration — and how to use each approach effectively based on real deployment automation experience.
Read more →Writing an Essay with AI: Codex vs Claude Code
Lessons from writing a 16-chapter essay with two AI coding tools — why prose is harder than code for AI, and what actually worked.
Read more →I Made a Product Demo Video Entirely with AI
How I used Claude Code to build a recording pipeline, edge-tts for voiceover, ffmpeg for assembly, and Gemini to QA the result — without touching a video editor.
Read more →Beyond RAG: Building a Recursive Language Model to Process 1M Tokens
How I built an RLM prototype that processes 71 arXiv papers (~1M tokens) without injecting them into the prompt, using out-of-core document analysis with Azure OpenAI tool calling.
Read more →Scaling your Agentic Workflow: Claude Code Agent Teams in tmux
How to orchestrate multiple AI agents working in parallel using Claude Code's new Agent Teams feature within a tmux session.
Read more →Is Claude a Co-Author? The Legal Debate No One Saw Coming
Analyzing the legal and ethical implications of 'Co-Authored-By: Claude' appearing in commits. Tool or author? The future of AI-generated code ownership.
Read more →TypeScript for AI Agents: From Friction to Flow with Sub-Agents
TypeScript provides essential guardrails for AI coding agents but creates friction with type errors. The solution isn't changing languages—it's changing workflows with specialized sub-agents.
Read more →Creating a Custom Skill for Claude Code: Automating Bilingual Blog Writing
How I built a Claude Code skill that standardizes and automates the creation of bilingual blog articles, ensuring consistency across EN/ES content.
Read more →Scaling Development with Parallel AI Agents
A workflow for running multiple Claude Code agents simultaneously on different features, using git worktree to manage branches and automating PR reviews.
Read more →Azure OpenAI's Content Filter: When Safety Theater Blocks Real Work
How Azure's content filter blocks legitimate browser automation tools based on word choice, not actual risk—and simple workarounds that expose the flaw.
Read more →Building an MCP Server for Bitbucket: Connecting LLMs to Your DevOps Workflow
How I built a comprehensive MCP server with 58 tools to manage Bitbucket repositories, pull requests, and pipelines through natural language.
Read more →