Goals

Active

• Enable real-time bidirectional conversation between human and AI #high
• Support natural interruption and course-correction mid-conversation #high
• Implement continuous speech transcription with minimal latency #high
• Create a turn-taking model that feels natural, not robotic #high
• Allow AI to listen even while speaking (bidirectional listening) #medium
• Build predictive response preparation that adapts as context arrives #medium

Future

• Create shared context window for visual collaboration #medium
• Support drag-and-drop of images, code, and documents #medium
• Enable AI to view and edit shared files #low
• Implement split view for multiple simultaneous files #low
• Integrate with existing tools (Ramble for ASR, Artifact Editor for context) #low

Non-Goals

• Full video conferencing - This project focuses on audio conversation, not video
• Multi-participant group conversations - Initial scope is 1:1 human-AI interaction
• Replacing traditional text chat - Complementary to, not replacement for text interfaces
• Real-time translation - Language translation is a separate concern
• Voice cloning or custom AI voices - Focus on conversation flow, not voice quality