The default assumes four hundred fifty leads each save zero point one five hours, or nine minutes. This assigns sixty seven point five staff hours saved each month. At twenty eight dollars per hour, gross labor savings are eighteen hundred ninety dollars. Subtracting the twelve hundred dollar retainer leaves six hundred ninety before setup. At one pixel per ten dollars the savings bar measures one hundred eighty nine, fee one hundred twenty and monthly remainder sixty nine. Annual gross savings less twelve retainers and thirty five hundred setup gives forty seven hundred eighty. None of these values measure actual dental practice traffic, staff time or successful calendar bookings. Payback uses a separate normalized benefit. It subtracts retainer minus an unexplained nine hundred dollar offset, so eighteen hundred ninety minus three hundred becomes fifteen hundred ninety. Setup divided by fifteen hundred ninety is about two point two months. Without that offset, setup divided by six hundred ninety is about five point one months. At one pixel per ten dollars the ordinary monthly benefit measures sixty nine, normalized benefit one hundred fifty nine and offset ninety. The formula also floors benefit at one hundred dollars, so it still reports finite payback when ordinary benefit is negative. The displayed payback is an illustrative assumption, not a verified client investment return. The script assigns eight hundred input tokens and two hundred fifty output tokens per lead, using saved rates of two point five and ten dollars per million. Four hundred fifty leads yield input cost ninety cents, output one dollar twelve point five cents and tool overhead forty three dollars sixty five cents. The sum is forty five dollars sixty seven point five cents, rounded to forty five dollars sixty eight cents. At five pixels per dollar input measures four point five, output five point six two five and tool overhead two hundred eighteen point two five. The initial markup says forty eight point six, but initialization recalculates. Cytoscape has a safe missing engine branch; native parameters and JSON proposal work locally, while diagram PNG needs the engine. Stress success is assigned, not measured.