The workflow I refuse to automate — and why that makes my AI builds better.
Problem: Every founder I meet wants to automate everything. But automating a judgment call doesn't save time — it scales bad decisions. The worst agents I've seen weren't underbuilt. They were trusted too far, too fast.
Build: Before I build, I run every workflow through three questions: Is it reversible? Does it look the same every time? What does a mistake cost? If a step fails any of them, it stays human — and the agent is built around it, with a clear handoff and an escalation rule.
Result: The automations I ship in my 5-day n8n x OpenAI sprints survive the first month, because the boundaries were decided on Day 1, not discovered in week 3. The runbook lists what the agent never touches, so the founder keeps control where it matters.
Offer: Tell me the one workflow you're unsure about automating. I'll tell you honestly whether it belongs in a sprint — and what should stay human.
Your first AI agent should ship in 5 days, not 5 months
Most founders I talk to have waited months for an AI agent that never ships.
Problem: scope grows every week — another integration, another edge case — while the manual work it was meant to replace keeps eating hours every day.
Build: I ship one working AI agent in 5 days with n8n and OpenAI. Day 1 is a 30-minute interview to draw the red lines — what the agent never touches. Then I build, test with an eval set, and hand over a runbook: what the agent does, what it never touches, and what to do when it fails.
Result: 3 AI agents now run in production for one client, with 1 paying subscriber live. The same eval-first pattern powers a PDF reader that returns 1 cited answer in seconds, and an iOS app with 5.0 stars on the App Store.
Offer: if one manual workflow eats your week — inbox, research, or ops — DM me and tell me what it is. I will tell you honestly whether an agent can fix it in 5 days, and what the eval would look like.
What I'm building right now: DUO, a functional prototype in active development.
It's a work and project management platform where AI agents take tasks in plain language, collaborate, and have their work checked by an independent agent. Projects keep memory across long sessions, tasks can be paused and resumed, and the owner approves what agents are allowed to do. Web UI plus an Android app.
Still in development, but it already shapes how I think about reliable AI automation.