"What running AI systems for my own business taught me about client work"
The best training I have had for building AI systems for other people is running them for myself — a multi-agent system that handles day-to-day operations for my own business, where every design shortcut eventually lands on my own desk. Building teaches you what these systems can do. Operating them teaches you what they cost.
Most of what I know about running LLM systems in production, I did not learn by building them for other people. I learnt it by being my own worst client. For a good while now, a multi-agent system has run a large slice of the day-to-day operations of my own business — triaging incoming email, extracting and routing information, drafting the routine responses, keeping the operational plumbing moving. When it breaks, my own work stalls. That arrangement has taught me more than any project I have delivered, because I am on both ends of every decision: I made the design shortcut, and I am the one it lands on three weeks later.
Have an AI feature stuck between demo and production?
The gap — reliability, evals, cost control, the plumbing that keeps it running unattended — is exactly the work I do. If that sounds familiar, a short conversation is usually enough to point you the right way.
Book a free consultation