Articles
One idea per article, pulled from the books and rewritten fresh, not chapter excerpts.
- AI-era delivery (2)
- Agents in practice (1)
- Quality & measurement in the AI era (2)
- People, career & AI (1)
AI-era delivery
Agile isn't dead. It's running on assumptions AI already broke.
Agile's values still hold, but many of its ceremonies were designed for a time when human effort was the bottleneck and feedback had to wait for a meeting. Now that AI has sped up how fast features get built, rituals like the sprint-end demo are turning into process kept alive for the organization rather than for the work. The practical fix is to name the assumption behind each ceremony and check whether it is still true.
Quality & measurement in the AI era
Quality used to mean "no bugs." It can't anymore.
When AI multiplies the amount of code a team produces, the old definition of quality, code that passed its tests and its review, can no longer be verified by people reading and testing it. DORA's 2024 research found that AI adoption, while it helps individual productivity, is associated with lower software delivery stability. Quality now has to be built in at the prompt, at the review of the whole output, and in layers of testing that used to be too expensive to run.
People, career & AI
Should anyone still study computer science? The honest answer.
For most people who want a career building software, a four-year computer science degree is now a weak investment: writing code has gone from the hardest part of the job to the easiest, and the industry will look very different by the time a new student graduates. Entry-level hiring has already shifted; Stanford research found employment of 22 to 25 year olds in the two most AI-exposed fifths of occupations fell about 11% between late 2022 and mid-2026. The degree still builds real thinking skills, but the skills that decide careers now are understanding customers, the business, and what is worth building.
Quality & measurement in the AI era
The metric you use to judge people is the reason your data lies
When a metric is used to evaluate people or teams, they start protecting the number instead of the reality it was meant to describe. In AI-assisted teams this shows up as healthy-looking burndown charts built on commitments sized for pre-AI capacity. The fix is to measure the system rather than the people, and to use metrics to ask questions instead of to hand out grades.
Agents in practice
I run a squad of AI agents. Here is what the work actually looks like.
An AI agent is a language model placed in a loop: it gets a goal, tools it is allowed to use, and permission to decide its next step based on what just happened. A squad is several agents, each owning a different kind of work. The clearest example is how this website is built: a writing agent, a critic, a builder and an automated quality gate, with one person approving. In my experience the limits are not intelligence but tokens, how precisely the job is described, and how requirements are defined before the squad ever starts.
AI-era delivery
AI made your team write code faster. Your delivery didn't notice.
AI coding tools have sharply increased how much code teams produce, but not how much they ship. CircleCI's analysis of more than 28 million workflows and an NBER study of AI coding agents both found output rising far faster than releases. The bottleneck moved to review and accountability, and teams need quality bars built for AI-sized output.