AI systems / built in public
Twenty agents on one framework. A research desk with a risk gate that can veto me. Two trading engines in testing. A pet brand that AI built in a week and runs every day. I document every build, including the parts that broke.
See the buildsNewsletter · weekly
One system teardown a week. What I built, what it cost, what broke, and the exact prompts that did the heavy lifting. No tool listicles.
Weekly. Unsubscribe in one click. Read past issues.
Real systems I run every day. Architecture, stack, and what broke, included.
It started as one assistant in a Telegram thread. It is now a framework of twenty agents on OpenClaw, each with its own job, its own rules and its own workspace, sharing one brain and eleven connectors. Tommy is the one I talk to. The rest report in.
Read the buildEvery morning dozens of scheduled AI jobs wake up across the cloud and a Mac mini: ad monitors, content runs, research desks, store upkeep. Before any of them does real work it asks a router which model to use, reads the same brain as everyone else, and follows one runbook. That is the bit nobody posts about, and it is the bit that makes the rest work.
Read the buildGordon used to write one letter on a Sunday. Now he runs a desk: thirteen specialist agents covering news, earnings, macro, charts, value and risk, a red team paid to disagree, thesis cards with kill criteria, and a risk gate that blocks any proposal that breaks the rules I set. It never trades. It argues.
Personal research tool. Not financial advice, not a recommendation to buy or sell anything.
Read the buildA crypto engine on OKX and a second bot that trades crypto on Binance and US stocks on Tiger Brokers. Code generates the trades, sizes them and manages every exit. A fast model picks one option from the menu or says hold. A slow model sets the mood once every few hours. Neither is allowed near the stop-loss.
Personal experiment. Not financial advice, not a recommendation to buy or sell anything.
Read the buildA real Shopify brand, live in 18 countries. AI researched the products, wrote every listing, set the prices, made the videos and the music, runs the social accounts, watches the ads twice a day and pulls the plug on losers, finds new products every Sunday and hands me a KPI review every Monday. I approve. It does the rest.
Read the buildEvery week a scheduled job researches the stories worth covering, checks each claim against at least five recent sources, makes the hero images and charts, and cuts every piece into a Medium copy, an X post, a LinkedIn version and an Instagram square. The store gets its own version for short video. Everything waits in one review sheet. Nothing ships without my yes.
Read the buildShort versions on X and LinkedIn, long versions on Medium, the files on GitHub, walkthroughs on YouTube.
Two open to everyone, the rest free for subscribers. One email unlocks everything.
Catalog
65+ free AI skills and MCP connectors for coding, design, e-commerce, social and more. Plus ten Shopify skills I built.
Browse itChecklist
The 12 things to lock down before you let an agent touch your email. Written after auditing my own and finding most of them open.
Read it freeCheat sheet Subscribers
Which model to use for which job, and what it costs. Stop paying frontier prices for formatting tasks.
Unlock with your emailPrompt pack Subscribers
The constraints I use to stop AI giving me hedged, agreeable rubbish.
Unlock with your emailSubscriber kit Subscribers
The operating manual template, twelve prompts and the launch checklist I used to build and run Tail Friends with AI.
Unlock with your emailI’m Cian. I’m Irish, I studied international business, and I’ve built my career in commercial and strategy roles across Ireland, China and Singapore, where I’ve lived for eight years: building markets, leading sales, and turning messy problems into plans people can actually execute. Working across three very different markets taught me to care about three things in any system: what it costs, who it serves, and whether it works when nobody is watching.
Over the last year I’ve pointed all of that at AI, and it got out of hand in the best way. I now run twenty agents on one framework, organised into teams. A fourteen-agent investment research desk with a risk gate that can veto my own ideas. Two trading engines in paper testing behind a promotion gate. A pet brand that AI built in a week and now runs every day, from product research to the ads. A content engine for two brands. And the mission-control layer that routes, remembers and audits dozens of scheduled jobs across all of it. The personal systems came first: a health coach, a renovation designer, an insurance brain and a smart home with no cloud.
None of these are demos. They are systems I depend on, so I write down how each one works, what it costs and what broke. The gap between what AI can do and what most businesses use it for is enormous, and the fastest way to close it is to watch someone do it for real.
More about meOffhand is the same approach, pointed at your business: agents that own a job, multi-agent workflows with one source of truth, store automation, content engines and GTM systems. Fixed scope, fixed price, and you own everything at the end. If you've read a build and thought "I need that", start there.
See the services or email