AI Systems
I Spent 18 Months Trying to Make AI Do Real Work — Here's What Happened
Custom GPTs, Gems, Projects, n8n pipelines — a year of paying for three AI subscriptions while still doing the heavy lifting myself. What finally changed.
The build log.
Honest notes from 18 months of trying to make AI do real work — not answer questions, not generate ideas, but take a brief and come back with something finished. The failures stay in. They're the useful part.
The articles below read in order as one story, or stand alone.
AI Systems
The Ralph Loop: How I Stopped Babysitting AI and Shipped a Real Product
Write a proper PRD, convert it to acceptance criteria, and tell the AI not to come back until everything passes. How a stupidly simple loop shipped a full whisky education platform in 15 evenings.
AI Systems
Running an Open-Source AI Agent at Home: The Hard Lessons
Broken updates, amnesiac memory, agents that lied about their work — and the architecture insight that made the whole painful experiment worth it.
AI Systems
My AI Team Has 7 Specialists and None of Them Are Human
How I run a multi-agent AI team on Hermes: specialist roles with their own memory and models, RouteIQ routing across nine models, and an escalation council.
Builds
One Rambled Voice Note, Two Working Apps: What Fable 5 Changed
I described an idea for fifteen minutes into Wispr Flow. The model planned, designed, wrote, built and tested it. How Family Taxi and RecipeMaster got built.
AI Systems
LLM Councils: Why I Make AI Models Argue Before I Trust Them
One model gives you a confident guess. Five models forced to debate give you defensible decisions for 25–50p. How my LLM council works and why platforms are now building it in.
AI Systems & Agents
How to Set Up AI Agents — Do This Diagnostic Before You Touch a Tool
Part 1 of a practical series on building an agent team that ships. The same discovery you should do before any system build — scope, time, money, skills, harness, and hardware — before you install anything.
AI Systems & Agents
How to Choose an AI Agent Setup — Path, Harness, and Models
Part 2 of a practical series on building an agent team that ships. Turn the Part 1 diagnostic into three decisions: which path you are on, which harness is home, and which model tier does which job.
AI Systems & Agents
How to Structure an AI Agent Team — Lanes, a Board, and Your First Agent Profile
Part 3 of a practical series on building an agent team that ships. Split the work into lanes, run it on a board, and write your first Agent Profile — so you stop being the human glue.
AI Systems & Agents
How to Stop an AI Agent Team Confidently Doing the Wrong Thing — Guardrails, QA, and Monitoring
Part 4 of a practical series on building an agent team that ships. Evidence gates, independent checks, tiered QA, and monitoring — so the team stops confidently doing the wrong thing, and you find out about failures from the system, not by surprise.
AI Systems & Agents
How to Give an AI Agent Team the Right Tools and Take It Live — Access, Workflows, and Go-Live
Part 5 of a practical series on building an agent team that ships. Tool access that matches the lane, a workflow that survives contact with reality, and the go-live checklist that switches the team on without switching your judgement off.
AI Systems & Agents
How to Grow an AI Agent Team Without Breaking What Already Works — Adding Agents, Scaling Processes, and Revisiting Old Decisions
Part 6 of a practical series on building an agent team that ships. When to add another agent, when to fix the one you have, how to add a second process without breaking the first, and the review cadence that stops old decisions rotting.
AI Systems & Agents
How to Run an AI Agent Team in the Real World — Hardware Gotchas, Model Behaviour, and the Truths Nobody Documents
Part 7 of a practical series on building an agent team that ships. The things nobody tells you: what actually happens on real hardware, how models behave when nobody is watching, and the operational truths that only show up after months of running agents.
AI Systems & Agents
How to Keep What an Agent Team Teaches You — Scars Turned Into Rules
Part 8 of a practical series on building an agent team that ships. The failures that cost time, money, and confidence — kept as scars, installed as rules, so you do not pay for the same lesson twice.