Guide
Automation pays off when you point it at the right tasks — and wastes money when you don't. The goal isn't to automate everything; it's to find the repetitive, rule-based work that drains hours and hand exactly that to a machine, while keeping people on the judgment calls. Here's a practical way to do it.
By Miguel Alejandro Hayes· Founder, Hayes Projects· with Claude (AI)
You can't automate a process you can't see. Break each process into individual tasks and, for every task, note how often it happens, how long it takes, and how many people touch it. The tasks that are frequent, time-consuming, and rule-based are your candidates.
This mapping is the whole game. Most teams are surprised how much time goes into a handful of repetitive steps once they're written down.
For each task, decide honestly: can a machine do it end-to-end (rule-based, high-volume, low judgment)? Does it need a person (empathy, negotiation, real judgment)? Or is it best done together — the machine drafts, a person approves? Automating the first bucket, leaving the second, and augmenting the third is where the real ROI lives.
Forcing automation onto human-judgment tasks is how automation projects fail. The value is in the split, not in blanket automation.
Don't automate the whole process on day one. Pick the single task with the best ratio of hours saved to implementation effort, automate that, measure the time it frees, and use that win to fund the next. Automation compounds — each freed hour is an hour redirected to work that grows the business.
An honest note: I wrote this with AI’s help. I used it to organize the research and tighten the prose, then reviewed, corrected and signed off on every line myself. Which is, more or less, the whole point: AI speeds up the draft — the judgment and the responsibility stay mine.
Free diagnostic
AutoMap maps your processes and shows what a bot can do, what needs a person, and what is better done together. No signup, in minutes.
Try AutoMap freeThe ones that are frequent, rule-based, and time-consuming. Our free AutoMap tool walks you through mapping your processes and scoring each task for automation potential in a few minutes.
Done right, it removes the repetitive work so your team spends time on judgment and relationships — the things that actually need a person. The goal is augmentation, not blanket replacement.

About the author
Economist and essayist turned developer. He founded Hayes Projects, a Miami venture studio and custom software lab, to build software that ships, scales and solves real problems.
Meet the founder→Co-author (AI)
Anthropic’s AI model. It helped organize the research, structure the draft and tighten the prose. Every claim was verified and signed off by a human.
Map your own processes free: AutoMap shows which tasks a robot can do, which need a person, and the hours you'd get back — no signup.
Try AutoMap free →Prefer email?
No obligation. No spam — just one useful, concrete reply.
We use first- and third-party cookies (Google Analytics and Google Ads) to measure traffic and improve the site. You can accept or reject them. Learn more