Hi, I'm Amen.
I build systems that listen and speak — voice AI engines, phone-line libraries, and the agent tooling that ties them together.
Writing
All notes →The anatomy of time-to-first-audio in a voice agent
From turn-commit to first audio: LLM ~59%, TTS ~28%, my orchestration ~14% — and one first token took 2.9 s. I was tuning the wrong layer.
Giving AI agents a real phone line
50 concurrent phone calls in userspace Go: p50 held at 20.00 ms at every load level — measured, not claimed. The tail is where it gets interesting.
Proof of life
2,551 commits across disclosed projects in the last year — 7 active this month.
26 weeks, one dot per week, sized by commits · most recently active first · counted only over projects listed on this page
Featured work
All work →xPhone
ShippingA phone line for your code — native SIP/RTP for Go and Rust, a bridge for every other language, and test tooling. 500,000+ production calls.
cadence
ShippingA provider-agnostic Go engine for low-latency STT→LLM→TTS voice pipelines — swap vendors with a config change. Owns the hot path.
Vocito
Private betaPush-to-talk voice typing for macOS — talk naturally, false starts and all, and it types the sentence you meant.
Shot Crisp
ShippingA macOS screenshot studio — capture, annotate, OCR, pin, and share; menu-bar native with an MCP bridge for agents.