Use AI responsibly to improve planning, analysis, documentation, forecasting, and management decisions.
Most project managers use AI like a smarter search box. The Prompt-to-System Method takes you from prompting to systems that run your sprint admin, in five layers, using skills you already have. The whole method, the two frameworks behind it, and the course that builds it with you.
Read article
Most lists of AI tools for Scrum Masters tell you what exists, not what is still on your desk. Sorted by what they read and what they hand back, the tools fall into three kinds. A one-page coverage map shows which of your reports they cover and which are still yours.
Read article
On a paid Claude plan, Claude Code and chat draw from one allowance, and Claude Code is almost always the heavy user. One command shows what is using yours, and the biggest drains have nothing to do with how many questions you ask.
Read article
Installing Claude Code is one line. People get stuck on the four decisions around it: which way in, which plan, which window and which update path. Answer those four and the install takes minutes, with or without a terminal.
Read article
Skills, MCP servers, plugins and subagents are not four names for one feature. Each does a different job, and picking by name instead of by job is why setups disappoint. One question sorts them, with a sprint report, a retro and a Jira board as the examples.
Read article
We built a seven-section course with Claude Code and kept a record of the failures. Facts it never checked, reviews that agreed with its mistakes, long instruction lists half done, gaps filled with guesses. Each one has a habit that catches it.
Read article
One sentence, five days of standup notes, one AI summary. It reads well and it is wrong in four places, all in two predictable ways. Here are the inputs, the output and the errors in full, plus the end-of-week check that catches them.
Read article
Most writing about automated status reports describes them. This one opens a public Claude skill and shows what it reads, how it decides red, amber and green, and what it cannot know. Three checks tell you whether any status report skill's colours mean anything for your project.
Read article
If being good at the job meant knowing the status, chasing updates and writing the Friday report, watching AI take those tasks can feel like losing the craft. This post takes that seriously, separates what we can prove from what we argue, and offers one exercise: the craft inventory.
Read article
Benchmark tables score the model, not the tool, and the best-known coding benchmark has been dropped by OpenAI. Four questions make a better recommendation: which plan, where your instructions live, whether you can read the code, and whether your organisation allows it.
Read article
Most AI pilots run by Scrum Masters and Agile PMs do not fail on the model. They fail because somebody pointed the thing at a decision instead of at a task. One test tells the two apart, and you can run it on your own backlog before you finish reading this sentence.
Read article
If you have ever stared at a Work Breakdown Structure and thought, “Great, now I need to write descriptions for all of this,” you are not alone. Breaking a project into manageable parts is one of the most important jobs in planning, but it is also one of the most repetitive.
Read article