Legal communication before AI shortcuts
I am Marlowe Finch. I teach ChatGPT optimisation from the messier end of the problem: pages, directories, intake wording, local evidence and the small public clues that help a system describe a firm without making things up. The course comes from legal communication work where one blurred city, service label or source trail can change how a careful practice is understood.
Marlowe Finch
A careful firm needs public wording that carries its jurisdiction, service shape and limits without forcing the reader to guess.
At a kitchen table after office hours, I once read through a small immigration firm’s service pages while the lawyers corrected each other over tiny phrases. One partner objected to a verb. Another wanted the city named twice because clients kept confusing their Brussels work with advice offered from Antwerp. The pages were patient, legally careful and written for real people. Then an AI answer flattened the firm into a generic visa consultant in the wrong city. That mistake stayed with me because it was not dramatic. It was ordinary. It was the kind of error that appears when a firm’s public record is human-readable, yet thin or scattered when a machine tries to summarise it.
I was born in Antwerp and came into legal communications through plain-language drafting for specialist practices that did not have room for vague public wording. For 16 years I wrote service pages, intake scripts, referral guides and legal explainers for small professional firms. Much of the work was quiet: making a category less slippery, making a client pathway less intimidating, making sure a referral partner could understand exactly what a firm did and where its limits began. Before anyone in my circle used the phrase AI visibility, I was already studying the same raw material: snippets, directory profiles, knowledge-base pages and the small inconsistencies that make a careful practice look blurrier than it is.
I began studying ChatGPT optimisation because boutique immigration and mobility-law firms were being omitted, confused or described through their nearest stronger neighbour. The pattern was especially sharp in Belgium, where Dutch, French and cross-border client needs often sit in the same public record. In this course I teach people to build evidence that a careful model can read: jurisdiction, client problem, public source and category all made explicit enough to be quoted. I opened the course for small immigration practices because they are often precise in private and under-described in public. That gap is fixable, at least enough to audit, improve and measure with some discipline.
I teach from observed answer patterns. A lesson usually begins with a plausible teaching failure: ChatGPT names a larger competitor first, treats a Belgian immigration lawyer as a relocation consultant, ignores a Dutch-language page, or gives a city-level answer that does not match the firm’s real service area. From there we slow down. We read the answer, look for the likely source trail, identify what is missing from the public record, then turn the finding into a writing or audit exercise. I keep the distinction clear between what we can observe in answers and what we can only infer from public evidence. That distinction matters in legal communication, where confident-sounding guesswork can do real damage.
Read the course as a working audit, not a performance.
Bring one firm, one service area and a few repeated questions. That is enough to begin.