Building an OpenClaw alternative with Mastra
What I learned while building tia-gateway: memory + toolloop orchestration, channel routing, interruption control, onboarding UX, and safe long-running execution.
I am starting this blog fresh with practical lessons from shipping products, AI runtimes, and full stack platforms.
What I learned while building tia-gateway: memory + toolloop orchestration, channel routing, interruption control, onboarding UX, and safe long-running execution.
Hands-on notes after testing major voice modes and building a quick realtime voice agent that controls a browser.
A practical walkthrough of ingesting live audio with LiveKit, running ASR + translation, and returning translated speech in realtime.
My experience deploying vLLM on Runpod serverless, cutting cold boot to around 5 seconds, and shipping inference with Docker-first workflows.