F14 -- Coaching & Debate Rehearsal

Provides AI-driven coaching for candidates including debate rehearsal, simulated press interviews, and town hall practice sessions.

Key Features

• Multiple coaching modes: Journalist interview simulation, debate rehearsal, and town hall Q&A practice.
• Configurable difficulty: Easy, medium, and hard settings control how aggressive the interviewer is.
• Topic focus: Specify topics for the session to practice specific policy areas.
• Conversation continuity: Multi-turn sessions where the AI adapts its follow-up questions based on the candidate's responses.
• In-memory sessions: Session state is maintained client-side (no database persistence required), keeping conversations private.
• Performance feedback: Post-session analysis of the candidate's performance.

Server Functions

Function	Endpoint	Description
start_session	coaching/start	Initialize a new coaching session
send_response	coaching/respond	Send a candidate response and get the next question
end_session	coaching/end	End session and get performance feedback

Coaching Modes

Mode	Simulation
journalist	Press conference or one-on-one interview with a journalist
debate	Debate stage with an opposing candidate
townhall	Voter questions at a town hall event

Difficulty Levels

Level	Behavior
easy	Straightforward questions, no follow-ups, friendly tone
medium	Probing questions, occasional follow-ups, neutral tone
hard	Aggressive questioning, persistent follow-ups, adversarial tone

Session Structure

pub struct CoachingSession {
    pub id: String,
    pub mode: String,          // "journalist", "debate", "townhall"
    pub topics: Vec<String>,
    pub difficulty: String,    // "easy", "medium", "hard"
    pub messages: Vec<CoachingMessage>,
    pub created_at: Option<String>,
}

Messages alternate between interviewer and candidate roles, building a natural conversation flow.

UI Components

• Coaching page (/coaching): Session configuration panel with mode, difficulty, and topic selectors.
• Chat interface: Real-time conversation view with the AI interviewer.
• Feedback panel: Post-session analysis with strengths, weaknesses, and improvement suggestions.

Database Tables

None -- sessions are maintained in-memory on the client for privacy.