Build an agent for one job
You tell me the job that eats your week. I build an AI agent that does it or speeds it up, around the tools you already use, with a person approving anything that goes out. We measure the time saved.
I'm Chris. I build and run AI systems that do real work, with a second model checking the work and a person saying yes before anything goes out. I build AI agents that take a job off your plate, and I look after them so they keep working.
You tell me the job that eats your week. I build an AI agent that does it or speeds it up, around the tools you already use, with a person approving anything that goes out. We measure the time saved.
Looking after two or three companies' AI setups each month: keeping the rules current, checking the work, fixing what drifts, and adding small builds when they're worth it.
Fully remote. I work with businesses wherever they are.
An MCP server with a web dashboard, written in Python. Claude, Grok, coding agents and ChatGPT (through its website) connect to it and are reminded of the same notes and rules at the start of each chat. It includes a setup guide, an undo history and a help wiki. It reminds them; it can't force them. Two testers, both friends, are trying it. Nobody pays for it yet.
About a dozen AI department agents research, draft and check. Each one has to pass an induction and starts on probation. A different model checks the work, and nothing is sent, spent or published without my yes.
Decisions, results and mistakes go into one record. When something breaks I write down what happened and what changed. Read the write-ups →
I don't yet have results for client companies with numbers attached. One tester has agreed to be my first case study. It will go on the work page when there's a real result. I haven't yet trained another company's team, and I haven't built in Copilot Studio, n8n or Make. I'd rather tell you that now than have you find out later.