— 000m · Work

Work

Reference architectures and case studies. Inspectable code, real measurements, and the failure modes you only find by running the thing on real calls.

— 000m Reference architectures

−450m Reference architecture

Low-latency voice agent: Conversation Relay + ElevenLabs + Deepgram + Claude

Reference architecture for a real-time voice AI agent on the phone. Twilio Conversation Relay as the transport, ElevenLabs and Deepgram on the audio I/O, Claude as the brain. Paired with a worked example in RecallIQ.

Status: Published View →

— 500m Case studies

−750m Case study

RecallIQ — voice AI study app on a $10/month box

PDF in, flashcards out, spaced repetition with two voice modes — one browser, one phone call. Built indie on Twilio Conversation Relay, ElevenLabs, Deepgram, and Claude. Live at study.samueleddy.com.

Status: Published View →