Writing

A series on running knowledge work in an agentic IDE, and short takes on enterprise AI in regulated finance.

One piece every two weeks. Each piece makes one claim a practitioner can test, shows a receipt from a real run, and states a limitation. Demos come with a recording, and some pieces ship a template you can clone. Pieces and takes post on Substack, each with a LinkedIn post, and the recordings are embedded in both posts. Templates and demo vaults go on GitHub.

A piece is listed as published once it is live, with its link.

The series

1. Six months of agents running my notes, measured. Next
   Point of view with three numbers from six months, a ledger reconciliation job and the rule for who may write which file.

Takes

Each short take on enterprise AI in regulated finance starts from a public event, with one opinion and one thing I would do differently. Takes post on Substack two a week until the first five are out, then one a week. A take is listed with its link once it is live.

1. October 21, 2026 A vendor just called its product an operational harness, which is good if we hold it to the word. Published
   Read it on Substack
2. The OCC has handed closure to your internal audit. Next

Elsewhere

Get started with ready-to-use AI frameworks, a Google Cloud article on AI Industry Bundles, coauthored with Fernando Rubbo, published August 28, 2025. Read the article