BravyOur approach
A little room before the real conversation.
How Bravy brings preparation, practice, and feedback together — and what the research can and cannot tell us about that approach.
Deutsche FassungKeep the context, not just the feeling.
In Bravy
Use your journal to record what happened, who was involved, and what you agreed. Journal features are free.
Why we built it this way
A note gives you something specific to return to when you prepare a conversation. This is a design choice, not a claim that keeping a journal has been shown to improve your conversations.
Give the conversation a starting point.
In Bravy
Plan a conversation in your journal or choose a scenario from the library. If you turn journal context into a practice, you choose and confirm what to share first.
Why we built it this way
We want practice to connect to situations you recognise. Research on behaviour modelling training suggests that practising trainee-generated scenarios can support transfer to work. That finding comes from training programmes, not from a study of Bravy.
Try it, then look at what you said.
In Bravy
Speak or type with an AI partner. After a practice long enough to analyse, review feedback on particular moments. When a moment is marked for retry, you can try a different response.
Why we built it this way
Responding to another turn gives you something to work with beyond a prepared opening line. A small study of a different AI practice tool found promising results for conflict strategies. It is a reason to explore this approach, not a guarantee about Bravy or your next conversation.
Two pieces of research worth reading
These studies inform our thinking. Neither evaluates Bravy. The summaries below stay within the findings reported in their abstracts; they are not a systematic review of the field.
Meta-analysis · 117 studies
Taylor, Russ-Eft & Chan (2005) · A meta-analytic review of behavior modeling training
The review found stronger effects on learning than on job behaviour or organisational results. Transfer was greatest under several conditions, including practice with scenarios generated by trainees and setting goals.
This examines behaviour modelling programmes with multiple training elements. It does not establish that a self-guided AI app produces the same effects, or isolate a Bravy feature as the cause.
Controlled study · 40 participants · CHI 2024
Shaikh and colleagues (2024) · Rehearsal — Simulating Conflict to Teach Conflict Resolution
After training with Rehearsal, participants used more cooperative and fewer escalating strategies in a subsequent conflict with a study partner than a group given lecture material on the same conflict theory.
A small study of a particular system and conflict framework. It is not evidence of long-term results, every workplace situation, or Bravy’s effectiveness.
What we do not promise
A simulation is not your colleague.
The AI does not know what another person thinks. Your real conversation can take a different turn, even if a practice went well.
Feedback is not an objective verdict.
AI feedback and scores can miss context or be wrong. Use them as prompts to reflect on what you said, not as a diagnosis of yourself or anyone else.
Research-informed is not product-validated.
The research above does not establish Bravy’s effectiveness. We do not promise a particular outcome, a promotion, or a change in someone else’s behaviour.
Practice has limits.
Bravy is not HR, legal, or clinical advice, therapy, or crisis support. For harassment, discrimination, threats, or other serious situations, seek appropriate qualified support rather than relying on a simulation.