<p>This article argues that the most consequential democratic question raised by artificial intelligence is not institutional efficiency. It is how representative systems can regain mechanisms for articulating, testing, and answering political claims. The synthetic chamber proposes a counter-institution to the privatisation of mediation occurring through polling systems, recommendation environments, campaign analytics, and conversational interfaces. We develop the concept of the synthetic chamber as a standing parliamentary organ of opinion-formation and reason-structuring designed to respond to the erosion of political party mediation. Its role is auxiliary in power but constitutional in effect, because what is institutionalised is a mechanism for processing and answering deliberative reasons within representative democracy. The proposal is grounded in debates on party decline, deliberative democracy, and representation, and treats mediation not as the replacement of human judgement, but as the use of multi-agent AI to structure, preserve, and relay the outputs of organised citizen deliberation. The synthetic chamber’s architecture is extrapolated from the research-creation setting of the, which made plurality, conflict, and traceability operable as AI deliberation. Rather than presenting a cure for democracy, the synthetic chamber offers a specific constitutional response to the contemporary crisis of mediation: a way of projecting organised reasons into representative politics in a form that preserves disagreement, compels answerability, and keeps algorithmic processing open to contestation.</p>