The last post made the agents fast. Fast got expensive. At peak, the work cost about $2.50 per PR, or $450 to $520 a day, at API-equivalent list price. Then I changed which model did which job, and how much each session read. Cost per story went from about $4.30...

Read on →

Two engineers worked on the same codebase with the same tool, Claude Code. Once the changes in this post were in place, I shipped about twice as many cards to users each day as the engineer who created the repo, about 37 against 18. Each one passed every check before...

Read on →

Two engineers worked on the same repo with the same AI tool, Claude Code. The engineer who created the repo had a strong setup. After I changed the engineering around the agents, I shipped about twice as many cards a day as they had. Then the cost per story fell...

Read on →

Here is the set of safety nets every team now needs, whatever its size. A story that read fine A team I know handed story grooming to an agent. They gave it a skill: a reusable set of instructions that taught the agent how the team writes user stories. It...

Read on →

I’ve spent the last year watching engineers and writers adopt these tools, and the same pattern keeps showing up. The first three months look transformative. The next nine look almost identical to the first three. The output keeps coming. The confidence keeps building. The skill stops moving. This isn’t a...

Read on →

A team I was talking to had built a code-review skill in Claude Code. It looked good: a careful prompt, a set of rules, and examples. Then it shipped a PR with a nested recursive loop inside another recursive loop. Cyclomatic complexity any static analyser (PMD, radon, ESLint’s complexity rule)...

Read on →

Production agentic systems decompose problems and use the right tool for each step. Most developers hand the AI the whole problem. That’s the gap. Teams building production AI workflows have developed patterns for making AI reliable. Developers using AI coding assistants like Claude Code, Cursor, or Copilot mostly haven’t adopted...

Read on →

Developers using AI estimate they’re 24% faster. A randomised controlled trial measured them at 19% slower. That’s from METR’s 2025 study. These were experienced open-source developers working on their own codebases with tools they chose. Their self-assessment was off by over 40 percentage points. If your perception of AI’s impact...

Read on →

Software engineering is changing. Again. I’ve spent the last two years applying AI across prototyping, internal tools, production systems, and team workflows. I’ve watched it produce an elegant solution in seconds, then confidently generate code calling APIs that don’t exist. I’ve seen it save hours on boilerplate and cost hours...

Read on →