Research turns up things that do not survive a second look, and things that are true but not ours to publish. Both are listed here, because a page that prints only the flattering half of its own evidence is a sales page, and this is supposed to be a record.
- How quickly those reviews arrived, or how often the business replies to them.
- The same read that produced the count produced both of those figures, and neither is published. A count is a position — the size of a reputation already built, which the business knows better than we do. How fast it is growing and how it is handled are operating facts about a company we are not working for, and this board does not print those about anybody.
- Anything about a state contractor licence, in either direction.
- The state's contractor search did not execute for us. Two attempts returned the search form itself rather than a result set, which is an instrument that did not run and not a search that came back empty — the difference matters, because an empty result and no result look identical in a note and mean opposite things. Their licensure is therefore unavailable to us, and unavailable is not a finding. In particular this page does not say they hold no licence, because we never got an answer to read.
- That this business is older than its live registration.
- Nine years is what a filing supports. A tenure figure on this board is the earliest Florida registration in a chain of entities linked by filings — not by a name, not by an address, and not by a person. A second company of the same name was filed in 2011 and is inactive, and it shares the name, the address and the manager. None of those is a filing. The two carry different federal numbers, neither record shows a merger, a conversion or a successor event, and neither carries any name history, so there is no chain to walk and this page reaches no further back than 2017.
- A founding date from anywhere other than a filing.
- The business publishes a number of its own for how long it has been working, and it is not restated here in any form — not quoted, not softened, not hinted at. What is published instead is the rule that produced the nine: the earliest Florida filing in a chain of entities linked by filings, and nothing else. That rule is the same on every row of this board, and it is the only defensible answer to a question two sources disagree about.
- How much of the AI assistants' answers this business holds.
- The survey read that axis and the number is not on this page. Part of the reading rests on a name merge rather than on an identity: a share of the cells credited to this business sit under a shorter string on one model, folded together by the reconciler, while the same answer set separately carries two other businesses with similar names. A merge that a reasonable person could draw differently is a judgment, and a judgment does not get published as a measurement. It is recorded in the research file where the reasoning can be checked.
- An ownership census of the businesses on this grid.
- Twelve dossiers on this board publish one: who owns the domains competing for a trade's searches, sorted into directories, franchise brands, private-equity groups and independents. That census is a reading over a list of web results. This trade's fifteen were read on a map grid instead, which returns businesses and classifies nobody's owner. Eighty-seven businesses take a top-three place somewhere on this grid and we know who owns none of them. Printing three zeros beside one large unclassified count would state three findings nobody made, and not knowing who owns something is not the same as knowing it is independent.
- How long this company's assets have been visible.
- Twelve dossiers here carry a web-archive first-capture date. No archive reading was taken for this business, so there is none on this page and none is inferred from its absence. The date this row carried until 3 August 2026 was a first capture of a different company's domain and it went out with the rest of that record rather than being re-pointed. The register date beside it is a separate observation and is not a substitute.
- That anything on this page can be compared with what this row published before.
- This row carried a figure of five of fifteen until 3 August 2026, and that figure was a count of how often one domain came back in a list of web results, for a different company. The figure on this page is zero of fifteen, and it is a count of map cells for this one. Two companies, two instruments, no common denominator. Nothing fell, nothing rose, and the older number is not restated anywhere on this page for exactly that reason.
- That the fifteen searches were read more than once.
- They were read once, on 2 August 2026 — twenty-five grid points each on the map side, and the answer side three runs each the following day. That is depth within a single collection rather than a second collection, and the two sides are a day apart rather than simultaneous. Nothing here says a place moved, held, or is where they usually sit.
- That every surface behind the answer readings was read.
- The instrument now grades five assistant surfaces; this record was read under the four it named at the time. Three of those four returned readings and the fourth, hand-run, carries none — so the answer side holds three of the five surfaces the instrument now grades. It is stated here because a short reading must not pass as a complete one, and because a reader is entitled to know the shape of the instrument before the number.