Install the macOS app
Open the Xcode project, run the app, and grant microphone, accessibility, and screen recording permissions.
live on clawd.x402.wtf
voice, vision, and solana data for your mac.
Beep Boop is a menu bar companion that sees your screen, listens on push-to-talk, speaks back, points at UI elements, and pulls live Solana context through a Cloudflare gateway.
OpenAIchecking
Heliuschecking
Birdeyechecking
Speechchecking
Talk
Hold control-option and say what you need.
See
The app sends screen captures through the gateway.
Point
The overlay flies to the relevant UI target.
how users actually get value in the first minute.
Keep onboarding direct: install, allow permissions, pick GPT 5.5, then ask for screen help or Solana wallet context.
cd beepboop/worker npx wrangler dev --port 8787 # Xcode app config ClawdGatewayBaseURL = https://clawd.x402.wtf # Run in Xcode open beepboop/leanring-buddy.xcodeproj
Choose GPT 5.5
The panel now defaults to OpenAI Responses through the deployed gateway, with Claude models still available as fallback.
Hold control-option
Ask for help with the thing on screen. The assistant can answer, speak, and point at the right place.
Ask about a wallet
Say a Solana address and the app injects live balance, Helius asset, and Birdeye price context into the model prompt.
what it can do for new users.
The product should be sold as a practical assistant, not a novelty. These are the first workflows to show.
screen-01
Explain the current app
Users can ask what they are looking at, where a control is, or what to do next inside a macOS workflow.
code-02
Help with build work
It can read Xcode, terminal output, docs, and UI state, then speak the next step while pointing at the right target.
sol-03
Read Solana context
Wallet balances, token assets, enhanced transaction history, and token price data are fetched by the gateway.
gateway console for demos and support.
Use this to prove the deployed gateway is alive before onboarding a new user or debugging their setup.
waiting for check
waiting for lookup
architecture adapted for the new public launch.
This is the old architecture visualization compressed into a scannable public explanation.
01Push to talkControl-option starts AVAudioEngine capture in the menu bar app.
02TranscribeAssemblyAI streams speech through the gateway token route.
03CaptureScreenCaptureKit captures all displays with pixel dimensions.
04ReasonOpenAI Responses reads transcript, images, and optional Solana context.
05SpeakElevenLabs returns low-latency voice audio through Cloudflare.
06PointThe overlay parses point tags and animates to the UI target.
how to get new users using it.
Lead with a short proof loop. People should see it point at something useful before they hear a long pitch.