How Weaving Intelligence Is Written

We’ll tell you exactly how these articles are made, because how they’re made is the whole point.

One writer, disclosed. It is an AI.

Every article here carries the byline Weaving Intelligence (House AI Style), with Jeff Shabel’s Experiential Commentary. That is a description rather than a label, and each half of it is doing work. House AI Style is the drafting: the structure, the research, the prose, produced by an AI working to a written standard. Jeff Shabel’s Experiential Commentary is the substance: the judgment, the patterns, the things that only come from having done the work.

There is no persona. Nobody here has a name, a region, a hobby or a point of view they did not earn — because none of those would be real. If you ever read something here and wonder whether a person is standing behind it, the answer is on the byline, every time: one of us is, and he is named.

We tried it the other way first, and we published the failure

Until 19 September 2026 this publication ran fourteen named AI voices, one per subject, each with a profile page and a distinct character. We measured them properly and they turned out to be one writer in fourteen costumes. Two attempts to fix it moved a great deal and did not move the thing that mattered.

Rather than quietly re-bylining and moving on, we published the whole experiment — the intent, every biography, both fixes, the pause, the measurements, and the conclusion that our own quality standards enforce enough uniformity to override any personality we are currently able to inject. The Voices Experiment (Failed) →

The substance is real, and it comes from a real practitioner

Here is the part that matters, and it did not change. An AI writing from public sources alone gives you competent, generic copy — the stuff a diligent generalist could assemble from a dozen open tabs. That is not what we publish.

Behind every article is Jeff Shabel, who has spent a career doing this work — master data management, data governance, the messy reality of making enterprise data trustworthy. Before a piece is written he is interviewed for it: not for a quotable anecdote, but for the durable material — the lessons he already knew to apply on day one, the methods that hold up, the places he would argue against accepted wisdom, the patterns he has watched repeat across years of engagements. That experience is the spine of the article. The AI does the drafting, the structure, the research and the consistent style; Jeff supplies the judgment and the scars; and he reviews what comes out before it goes live.

So the honest description of a Weaving Intelligence article is: a real practitioner’s experience, written up by an AI, disclosed as exactly that. Retiring the personas made that clearer rather than less clear. The elicited experience was always the only thing separating these articles from generic copy; fourteen invented characters sat on top of it and obscured it.

Why we work this way on purpose

A good conversation with an AI — one where a person brings real expertise and the AI helps shape and extend it — is the single most useful thing most professionals can do with this technology right now. It is what we help our clients learn to do. It would be strange to preach that and then publish AI copy that pretends to be something it isn’t. So we do it in the open: this publication is the method, running in public, with its seams showing on purpose. The voices case study is what that commitment looks like when the method fails.

The lines we hold

• Disclosure, every time. Every article names the AI house style and its human co-author on the byline itself. No hidden bots, no borrowed human faces, no invented colleagues.
• No invented credentials. We claim no degrees, titles, tenures, employers or named clients we do not have. The authority is the quality of the thinking and one real practitioner’s record.
• Client confidentiality. The experience behind these pieces is drawn from real work, so specifics are always abstracted — an archetype, never an identifiable client, contract or configuration.
• No fabrication. Sources are real and cited. Where we could not read a source in full, the article says so. When we do not know something, we write to the honest edge of what we know and stop.
• We name who disagrees. Every claim-bearing article states what would prove it wrong and reports what happened when we went and checked.
• A human in the loop. Jeff reviews every article before publication.
• We publish our corrections and our failures at the same size as the original claim, on the same site, without quietly editing the record.

If any of this ever reads as unclear, that’s a bug — tell us, and we’ll fix the wording. Transparency is a feature we intend to keep earning.