Method
How the research is done, and who is answerable for it.
Every report is built from primary sources, and every figure carries its source, its date and a label saying how much weight it will bear. This page sets out how that is done, what we will not do, and who answers for it.
Values
Four commitments you can hold us to.
Each of these is a promise about how we behave, not a claim about how clever we are. Every one of them can be checked, and every one of them costs us something.
Integrity
We publish what the evidence supports, not what sells best. A figure we cannot confirm at source is removed rather than softened, even when it was the most quotable number we had.
Impartiality
We are not paid by the companies we write about, and no vendor sees a report before publication. Where sources disagree, we show both and let you judge.
Show the working
Every figure carries its source and its date, and a label saying how much weight it will bear. Research published by a vendor about its own product is flagged as such at the point of use.
Answerable
A named person holds editorial responsibility for every report. If we get something wrong we correct it, date the correction, and tell everyone who has already bought that report.
Curiosity, systems thinking, judgement and continuous experimentation matter a great deal to this work, but they belong in the writing as an argument about what makes AI adoption succeed, not here as a claim about ourselves. You cannot check whether we have good judgement. You can check whether a figure has a source.
The method
How a report is checked.
Sources are gathered, summarised and cross-checked with AI assistance under a fixed method. Every figure is then checked against its original source before it appears, and a figure that cannot be confirmed at source is removed rather than softened. Nothing is published on the strength of an AI summary alone.
Allen Chen holds editorial responsibility for every report: for the selection of sources, for every figure and claim, and for its corrections. A research log recording sources, checks and corrections is kept for each report and each version.
We are a small publisher. Where a report says "we", that means Allen Chen and the tools he directs – not a research department.
The operating model behind that sentence is set out in full, and dated, at how this business runs: four agents, what each may settle on its own, and what comes to a person.
Interests declared
Allen Chen also provides independent AI advisory services, and therefore has an interest in the subject of the AI in TIC series. Deep Dive Reports receives no payment, commission or other consideration from any product or vendor named in a report, and no vendor is given sight of a report before publication.
Naming other organisations
We name a company only where a claim about it can be traced to a verifiable source. Where the only material available is a company's own unaudited communication, we describe the sector rather than name firms. Products are named only because they are in current use, are listed alphabetically, and none is assessed, ranked or endorsed.
We do not compare, score or rate other organisations. That is a deliberate limit, and it costs us something: it rules out the competitor-comparison work that clients ask for. We would rather forgo it than publish a judgement about a company that has no way to answer back. The full rule is in the Disclaimer.
Corrections
Found a mistake? Tell us at allen@deepdivereports.com. We will correct it, record the date of the correction, and if the change is material we will tell everyone who has already bought that report. Every report carries a version number, so we can always say which document you received.