Judge map

We made friction the product.

Unread takes a hard position: the right AI for reading should not make every text easier. It should make pretending harder.

30%

Interpretation & stance

We built Unread to study productive friction because reading tools usually remove the exact struggle that proves comprehension.

25%

Creative leap

The product refuses the obvious AI-summary answer. It uses AI as resistance: a gatekeeper that says, 'not clear yet.'

20%

Craft & execution

Saved rooms, streaming AI feedback, locked sections, presence, reports, public posts, and author analytics all serve one idea.

15%

Communication

A stranger should understand it in one minute: paste text, answer the claim, get blocked when you fake it.

10%

Conviction

Unread is willing to be annoying. That discomfort is the stance, not a side effect.

Demo path

Do this live.

1

Open the seeded post: The meeting after nobody read.

2

Show that only the first section is visible.

3

Submit a vague answer and let the AI push back.

4

Submit a precise answer and unlock the next section.

5

Open the dashboard analytics to show the weak checkpoint.