[ METHODOLOGY ]
The questions are real. Every board is a question New York buyers and sellers actually type into AI engines — selected using search-demand data, phrased the way people phrase them.
The sessions are clean. We ask through the engines' public APIs: stateless sessions with no account, no history, no location profile, no memory. The same blank slate every time. What the engine says to a clean session is the closest thing there is to what it "really thinks."
The answers are published verbatim. Names, order, firms, and the engine's own reasoning notes appear exactly as answered. We add nothing and remove nothing. Each board shows the date of the pull it reflects, and every prior pull is archived.
The engine roster today is ChatGPT (OpenAI API with live web search). Gemini and Perplexity boards are in the pipeline; each board is labeled with the engine that produced it.
Nobody can buy a seat. Check The Scoreboard is published by Position Strategy Group, a firm that sells AI-visibility consulting. That relationship is disclosed here and in the footer of every page, and it comes with a hard rule: clients get no lift, no edits, and no seats. If a sponsored feature ever appears on this site, it will be labeled as one. The rankings never are.
What a board is not. A board records how AI engines answer a question on a given date. It is not our opinion of any agent, and engines can be wrong — that gap is, in fact, the story this site exists to document.