This is a YouTube thumbnail where, on the right side, there is an image of a person wearing a white mask being interviewed.
Using this aesthetic, I want to share the information architecture for the thumbnail with the AI and have it create the thumbnail, so please think of a prompt.

The reference image is only for the masked person on the right ({argument name="character_name_en" default="Eru"}); use that as the reference, and make the layout, text size, and design so that the click-through rate is maximized.

The target audience is men in their {argument name="target_age_range_en" default="late 20s to 30s"} who struggle with communication skills.

# Thumbnail information architecture
<Left side>
Three layers of conversation (small)
Express them as an inverted triangle
Divide into three colors
In reality, from top to bottom, the following three (blur the text):
・Abstract
・Facts
・Concrete

<Right side>
{argument name="character_name_en_again" default="Eru"} (the masked person)
“This is all there is to conversation” (put this in a speech bubble)

<Bottom>
“With this,
conversations go on forever” (stretch across the full width)