Ask a founder what they'd like to automate and they'll describe something clever: an AI that writes proposals, a bot that qualifies leads, a dashboard that predicts churn. Ask them where their week actually goes, and you'll hear something different: copying data between tools, chasing invoices, posting the same update in three places, formatting the same report.
That gap is where most automation budgets die. The clever 20% is hard to build, fragile in production, and rarely trusted by the team. The boring 80% is none of those things.
The three-question filter
Before automating anything, I run it through three questions. Does it happen at least weekly? If not, the maintenance will cost more than the manual work. Are the rules already written down — or could a new hire do it from a checklist? If a human needs judgment, automate around the judgment, not through it. Does anyone notice when it's done wrong? If yes, keep a human in the loop for the final step.
Anything that passes all three is a candidate. Then rank the candidates by hours saved per month and start at the top. That's the whole method — deliberately unexciting, reliably profitable.
Automation isn't about replacing people. It's about deleting the work nobody should have been doing by hand.
Where AI actually fits
AI belongs in the messy middle of a workflow — drafting, classifying, summarizing — with structured automation on both sides of it. At Veziro, the AI drafts the marketing content, but the scheduling, publishing, and reporting around it are plain, deterministic workflows. Boring in, boring out. That's why it runs every day without anyone touching it.
Automate the boring 80% first. The interesting 20% will still be there — and by then you'll have the time and the data to do it properly.