A manifesto for better conversations

Your AI is
lying to you.

ChatGPT doesn't analyze your messages. It agrees with whatever you're already thinking.

Get actual analysisFree to try - No card needed
The problem with AI feedback

You ask:

"Does this sound too passive-aggressive?"

ChatGPT says:

"No, your message sounds perfectly fine! It's clear and direct. Great job!"

Then you ask:

"Actually, is it too aggressive?"

ChatGPT says:

"Not at all! Your tone is perfectly reasonable. You're being assertive."

It's not thinking about your message.
It's agreeing with you.

A different approach

Four perspectives that
argue with each other.

S

Signal

What does it literally say?

"You used 4 hedging words. You said 'just' twice and apologized once."

O

Opportunity

What leverage do you have?

"They need something from you too. You're not asking for a favor."

A

Affect

How will it feel to receive?

"The excessive politeness might read as desperate or low-confidence."

R

Risk

What could go wrong?

"You gave them 3 easy outs. They'll probably take one."

"Put my job complaint through it. It said I wasn't venting - I was rehearsing a resignation I wasn't ready to give."

- Marcus, 24, Software engineer

Real conversations, real stakes

For the messages you'd never
bring to therapy.

The text to your ex

Before you send that 2am message you've rewritten 6 times.

The salary ask

Stop apologizing for wanting to be paid what you're worth.

The roommate confrontation

Address the dishes without starting a cold war.

The cold DM

Reach out to that person without sounding desperate.

The work email

Set boundaries with your boss without getting fired.

The apology

Make it count. No more "sorry if you felt..."

$4/month.
Less than a latte.

For the conversations that actually matter.
The ones you overthink at 2am.

Try it free
✓No credit card✓Cancel anytime✓Instant access
4Angles

Anti-sycophancy AI for real conversations.