How to use it
What this calculator actually measures
Automation ROI is the difference between the labor cost you stop paying and the new costs you start paying (AI tokens, platform, and the human review everyone forgets). This tool computes the monthly net, the return on operating cost, and — the number decision-makers care about — how many months until the build pays for itself.
How to estimate the AI cost per task
One automated task usually means one or a few LLM calls. Model your task's tokens in our AI API cost calculator — its "cost per 1,000 requests" output divided by 1,000 is your cost per task. Typical document or email tasks land between $0.01 and $0.10 on mini-tier models.
The two numbers people get wrong
- Automatable share — no automation handles 100% of cases. Estimate the share of routine cases and let the rest escalate to humans; 60–80% is a healthy target.
- Review minutes — someone will check outputs, at least by sampling. Even 1 minute per task across thousands of tasks is real payroll; put it in the math, not under the rug.
When NOT to automate
If the task is rare, changes constantly, or the cost of a wrong result is high (legal, medical, irreversible actions), the review burden eats the savings. A negative result in this calculator is a useful answer — it stops you from spending $2,000 building something that saves $40/month.
Worked example
A team triages 500 support emails/month at 15 minutes each ($25/h). Automating 70% with AI ($0.05/task, $50/month platform, 1 minute of review each, $2,000 build): the automation replaces ~$2,188/month of labor, costs ~$213/month to run, and nets ~$1,974/month — payback in about a month, ~82 human hours freed. Those are the defaults loaded above: replace them with your task.