Cameras have previously been capable of just saving pixels for someone else to watch − the computers themselves could not understand what was in the image. Now though, cameras can recognise faces, emotion, gender and age.

Bars can currently analyse sales, but the patrons are coming for more than just alcohol; they’re coming to be part of a curated social space. Since our societies use visual communication so heavily, the use of cameras allows us to capture so much that the POS was not telling us.

• How long does each patron stay in the bar?
• How often does each patron return to one of your bars around the city?
• How many new patrons are visiting your bar for the special event you organised?
• Did a celebrity just enter your bar?
• How often is a patron smiling tonight, compared to their regular behaviour?
• How old is the typical crowd at each of your bars?
• How old are your wine drinkers vs beer drinkers?
• How does the gender distribution vary throughout the night?
• Are there any patrons whose presence affects the gender distribution?

Bars can currently analyse sales, but the patrons are coming for more than just alcohol; they’re coming to be part of a curated society. Since our societies use visual communication so heavily, the use of cameras allows us to capture so much that the POS was not telling us.

By being able to match visual data with other signals, Tanuki can learn new visual cues:

• What does it look like when someone is going to order a drink in a minute?
• What does it look like when a glass is going to be broken in a minute?
• What does it look like when a date is not going to last?
• Who left this glass here?
• What do each of your menu items look like?
• Are your staff acting abnormally today?
• What is each patron’s blood-alcohol concentration?
• What is each patron’s heart rate?