~/temisangerrard/writing $ ▋
Technical articles from production AI systems. No tutorials. No hype. Decisions, rationale, and what broke.
2026-03-21
~8 min read
L01
L02
L03
L04
L05
ARCHITECTURE
+ CASE STUDY
444 commits, 4 production apps, real USDC on Base mainnet, autonomous bots with no human in the loop.
Here's what breaks when agents are live, stateful, and holding other people's money — and the five infrastructure layers that don't appear in any LLM tutorial.
Read article
arrow_forward
2026-03-28
~6 min read
L03
L04
AGENTS
+ GTM AUTOMATION
45 minutes of setup, 15 hours/week saved, 2 minutes of human input. How I automated competitive intelligence,
backlink prospecting, content strategy, and partnership research — then documented what broke.
Read article
arrow_forward
2026-03-31
~7 min read
L01
L04
L05
INFRASTRUCTURE
+ COST BREAKDOWN
A real Fly.io invoice for Arena Play — $110.75/month to run a production AI product on Base mainnet.
Redis is 60% of infra cost. Compute is 25%. The gap between "talking about AI" and "running AI" is visible in your billing dashboard.
Read article
arrow_forward
2026-04-03
~9 min read
L01
L02
L03
L04
L05
AI PRODUCT
+ CASE STUDY
303 commits, 486 TypeScript files, 28 database migrations. The architecture where the app decides what to cook
and the LLM only writes the recipe. Cultural food knowledge encoded deterministically. Defensive parsing as a survival skill.
38% of commits are production hardening nobody writes articles about.
Read article
arrow_forward
// next article in progress