Writing
Notes from inside the work
Occasional, specific and written after the fact — no schedule, and nothing published for the sake of publishing.
What we write about
Four recurring subjects, and one rule: if it is not something we have actually done, it does not get published.
What is worth automating, and what is not
Most of the value is in a handful of processes, and a fair amount of harm is in the ones that look automatable but carry a judgement call. Telling them apart is most of the work.
Making AI-assisted work auditable
Systems that cite their sources, show their reasoning and stay reversible. Not because it is fashionable, but because nobody can defend a decision they cannot explain.
Integration without a new platform
How to make the tools a business already runs agree with one another, and why the answer is almost never a fourth system.
What went wrong on real engagements
Anonymised, specific, and including the parts that did not work. A case study with no failures in it is marketing.
By email instead
The same pieces, sent when they are published. We confirm the address first and unsubscribing is one click.
Would rather talk than read?
Thirty minutes, no preparation needed. Vanessa Carmin will ask the questions.