Features
Screen-aware AI that sees your current application and context
Voice input via push-to-talk with AssemblyAI transcription
Speech output via ElevenLabs text-to-speech
Step-by-step walkthroughs for design, video editing, and coding
Cursor-adjacent menu bar interface that doesn't block workspace
AI can point at screen elements using [POINT] tags in responses
Powered by Claude (Anthropic) for AI reasoning
Open source under MIT license – fork, modify, reuse freely
Cloudflare Worker proxy for secure API key handling
Real-time audio streaming over WebSocket for low latency
Streaming SSE responses from Claude
Requires macOS 14.2+ and Xcode 15+ for building
Runs on both Apple Silicon and Intel Macs
Optional local development with wrangler dev
AI Agents for automated workflow execution
Flow Builder low-code automation design tool
Sidekick natural language AI assistant
Flow AI in-platform automation assistant
Voice AI for patient communications
Automated copay estimation with RTE verification
Care gap outreach and self-scheduling
Automated referral management and leakage reduction
Automated appeal letter generation
Real-time eligibility verification
Patient intake and registration automation
Order transcription automation
Denial prevention workflows
Pre-procedure instructions via Voice AI Agent
