Productivity & daily use

AI Task Automation Checker

Seven questions decide whether a task is worth automating. The score, the hours saved and the payback are all arithmetic you can check.

You do this task at least weekly, the same wayOne-off tasks rarely repay the setup time.The inputs are text, numbers or filesPhone calls and hallway context do not automate.You could write the rules down for a new hireIf you cannot explain it, you cannot delegate it.A wrong answer costs minutes, not moneyHigh-stakes output needs a human signature.No customer or regulated data leaves your systemsOtherwise you need approved tooling first.Nothing physical has to happenAnything requiring hands stays manual.You can verify the output in under a minuteUnverifiable output creates hidden rework.

Minutes per runRuns per weekSetup hours

Automation score

84 / 100

Automate this now

Time spent today: 2.5 h / week
Hours saved: 1.6 h / week
Hours saved per year: 77 h
Setup pays back in: 5 weeks

What to do next

1. Write the task out as numbered steps with one real example of input and output.
2. Automate the single slowest step first and keep a human approval click at the end.
3. Log every run for two weeks and compare against the manual baseline.

How the score is built

Each question carries a fixed weight, heaviest on the three things that actually predict a successful automation: the task repeats, the inputs are already text or numbers, and the rules are explicit enough to hand to a new hire. A partial answer scores half.

Savings are deliberately conservative. Above seventy-five you keep about sixty-five percent of the time; in the middle band forty percent, because a human still reviews every run; at the bottom fifteen percent, which is roughly the value of a faster first draft and nothing more.

Payback divides your setup hours by the weekly saving. Anything over twelve weeks usually gets abandoned before it repays, so treat that as a signal to shrink the scope rather than push harder.