A personal log of what happens when a non-engineer decides to take AI seriously — from first hello to running a multi-agent research system.
First message to my new AI assistant my AI assistant. Within an hour we had Gmail, Google Docs, Drive, and Sheets all connected — and a voice transcription pipeline running through Gemini.
Five agents. One pipeline. A daily 6 AM briefing with AI-curated stock picks and tech news delivered before market open. Here's how we built it in a single session.
Five specialized agents, one pipeline, and a switch to Tavily for cleaner news. A breakdown of who each agent is, what they do, and why separating the roles makes the whole system sharper.
What I learned running an AI research swarm every morning — which picks worked, which didn't, and how the agents got sharper as they built up memory over time.
A friend's wedding site from 2004 was rotting on old GoDaddy hosting. We rebuilt it, recovered missing images, and deployed it to Cloudflare Pages — all in an afternoon.
Moving two domains from GoDaddy to Cloudflare looked simple. Then came the CNAME flattening, the stale router cache, the SSL pending state, and the Error 522. Every problem was a lesson.
A car detailing business needed a premium website. The AI designed it, I deployed it, and we had it live at a custom domain within the hour — zero traditional web hosting involved.
I watched a video about agents arguing about stocks. Then I realized the same pattern works for hiring decisions, purchases, architecture choices — anywhere you need a second opinion that actually pushes back.
Claude + Gemini 2.5 Flash + GPT-4o-mini, in parallel, before every reply. Here's the architecture, the cost breakdown (~$0.0003 per check), and why I decided the latency was worth it.
The meta-post. I asked my AI assistant to build a blog about our journey. She read 17 conversation transcripts, extracted the key moments, wrote the posts, deployed the site, and configured the DNS. I picked the domain name.