Thoughts
Long-form articles and guides on AI, voice technology, product development, and more.
Create Your Own AI Voice Agent
A complete guide to building a custom AI voice agent with Hume EVI, Claude, and Railway. Learn the architecture behind RyBot including code samples and implementation details.
$3.99 — Unlock
The LLM Council: Six AI Models Review Everything I Build, via One OpenRouter API
I took Karpathy's llm-council idea, cut it to the one stage that adds signal, and made it a daily workflow: six frontier models anonymously critique my drafts through one OpenRouter API, and Claude clusters the findings.
3D Printing The Met: A Guide to the Museum's Free CC0 Scans
The Met released ~140 high-res 3D scans of its collection under CC0. How to find them, download them, convert them to STL, and print them at home, with three interactive models to explore.
Building Trading Buddy: An Open-Source Autonomous Trading Platform
The story behind Trading Buddy - why I built an autonomous trading platform with multi-agent architecture, and how you can use it too.
Fast Enough to Learn
Why teams that learn to deliver quickly end up building better products. Lessons from 15 years of early-stage product development on iteration, feedback, and momentum.
Crafting a Voice Agent with Hume AI and Claude Code
An introduction to building voice AI agents with Hume EVI and Claude. Learn why this combination works and what's possible.
Securing Voice AI: Preventing Clone Attacks & Protecting Users
A practical guide to voice AI security. Learn how to protect your voice agents from clone attacks, prompt injection, and social engineering while maintaining user privacy.
Building AI Memory with Supermemory
How I built persistent memory for RyBot using Supermemory. Covers user-specific memories, a shared knowledge base for personality emulation, and an interactive knowledge graph visualization.
Voice Clone Security Guidelines
A comprehensive security framework for AI voice clones to prevent fraud, identity theft, and misuse. Version 2.0 covers anti-verification protections, channel constraints, crisis handling, and application-level guards.