About

This is a reference work for people running enterprise AI transformation inside large organisations. It is organised around a numbered canon of patterns rather than a stream of posts, because the problems repeat and the good answers change slowly.

How it is maintained

Entries are revised when we learn we were wrong, and every entry carries the date of its last substantive change. Behind that, an automated job re-reads the citations each week, refreshes the measured data, and tells the editor which entries have gone too long without somebody checking them against current practice.

That queue is deliberately not published. Maintenance is something a reference work owes its reader, not a status report to put in front of them, and an entry either currently holds or it should not be here.

Where the words live

Entry and essay text lives in the source repository and reaches this site through a reviewed commit. The database holds maintenance metadata and measured instrument data, and nothing else. The automated job that refreshes instruments and records freshness checks has no commit access and no database grant that would let it change what an entry says, publish, retire, or renumber anything. That is enforced by column-level permissions rather than by policy, and we test it.

Corrections policy

Corrections are recorded as corrections, kept separate from ordinary revisions, because “we developed this argument” and “we were wrong” are different statements and folding one into the other is a way of avoiding the second. Where a correction changes what an entry concludes, it is stated in the entry.

• Factual errors are corrected as soon as we can confirm the error, and the correction is logged with what was wrong.
• Essays are not rewritten. An essay stands as published on its date. A correction to an essay is published as a correction rather than edited in silently.
• Instrument data is republished in full when a method changes, and the method change is described. We do not quietly restate a series.
• If a claim cannot be sourced to a primary record, it comes out. A number we cannot show you how to reproduce is not evidence.

To report an error, write to jindal.rahul@gmail.com. Tell us what is wrong and, if you can, what the right answer is and how you know.

What it costs

The canon, the lexicon, the instruments and the raw data behind them are free, and stay free. Membership of the room is free permanently, and that is a design decision rather than an introductory price. The moment money moves between the host and the members, the host becomes a supplier, and a room of suppliers and customers is not a room of peers.

Where money does have to move, it moves at cost. A printed edition has to be printed and posted. People in a room need the room. Anything like that is priced to cover what it actually costs and nothing further, and the arithmetic is published beside it, in the same way a number in an instrument is published with its method.

There is no sponsorship, no advertising, no affiliate arrangement and no paid placement anywhere here, in any entry, instrument or index. That does not change if any of this becomes popular. If you ever find a figure on this site that does not reconcile, or a price that clears a margin, that is an error under the corrections policy and it should be reported like any other.

What this is not

• Not a consultancy. There is no engagement behind the door, and nothing here is a route to one.
• Not vendor-funded. No sponsorship, no affiliate arrangements, no paid placement in any entry, instrument or index.
• Not an employer publication. See below.
• Not a newsletter with a reference section attached. The reference is the thing.

Disclosure

Published in a personal capacity. Nothing here represents the views of any employer, and no non-public information is used.

The editor works in enterprise technology, and writes separately in a first-person capacity at byrxj.com. The two publications share an editor and nothing else. No text is republished between them. Where a piece here relates to one there, it is linked rather than reproduced.

Nothing published here uses non-public information from any employer, customer or member of the room. Where an entry describes a pattern observed across organisations, it is generalised to the point where no organisation is identifiable, and the room operates under Chatham House rules permanently.