ABOUT BRAMBLE

Bramble challenges thinking and suggests experiments. It tells you when to talk to a real person.

We only use paid model providers that don't train on your conversations.

Every response follows a framework — validate your situation, check for distorted thinking, offer multiple strategies with scripts you can actually use, and suggest one concrete next step.

Before generating a response, Bramble checks whether it has enough context about the relationship, what you've tried, and what you want. If not, it asks.

Responses are checked for mind-reading, unsafe suggestions, and quality. If something looks off, it automatically retries.

Within a conversation, Bramble remembers what you said so follow-ups build on context. Paid accounts get cross-session memory — Bramble learns your patterns over time and avoids repeating advice that didn't work.

The model focuses on observable behavior and concrete experiments, not abstract advice. "Say this sentence Thursday" beats "try to be more assertive."