Whether an AI agent pays off can be estimated with a simple method. You do not need a spreadsheet with twenty tabs, just honest answers to three questions. This article shows the formula, works through an example with clearly labelled assumptions and explains how to tell after a few months whether the numbers hold up.
The formula in three steps
Step 1: How many hours of routine work come up each month? This means recurring work that follows a fixed pattern: copying quotes from emails, confirming appointments, syncing data between programs, answering standard requests.
Step 2: What does one of those hours cost, fully loaded? Not just gross pay, but also employer contributions, workplace and time off. If you are self-employed, use what you could otherwise bill in that time.
Step 3: What share can the agent take over? It is rarely one hundred percent. Exceptions, questions and approvals stay with people.
This gives you the gross monthly saving:
hours of routine × hourly cost × share handed off
From that, subtract the time newly needed for reviewing and approving. The result is the net saving. Against it stand the one-time setup and the monthly support fee.
A worked example
All figures in this example are assumptions for illustration. They do not come from a customer project and are not a price quote.
| Step | Assumption | Result |
|---|---|---|
| Routine per month | 60 hours | |
| Fully loaded hourly cost | €50 | €3,000 routine cost |
| Share handed off | 50% | €1,500 gross saving |
| New review time | 5 hours × €50 | − €250 |
| Net saving per month | €1,250 | |
| Setup (one-time) | €6,000 | |
| Monthly fee | M (as quoted) |
We deliberately keep the monthly fee as a variable M. It depends on scope and is only set in the quote.
Break-even
The point at which the setup has paid for itself is calculated like this:
months to break-even = setup ÷ (net saving − monthly fee)
In the example: €6,000 ÷ (€1,250 − M).
You can also read it the other way round and see which monthly fee fits your goal. If the setup should have paid for itself after twelve months, the difference must be at least €500. In this example the monthly fee could therefore be at most €750. With a 24-month target it could be at most €1,000.
If the monthly fee is higher than the net saving, the agent does not pay off through time savings alone. Then other effects have to make the difference, such as faster replies to customers, fewer errors or software subscriptions you can cancel.
What distorts the calculation
Three things are often overlooked:
Ramp-up. In the first weeks an agent usually takes over less than planned, because special cases are still being refined. Be cautious for the first one or two months.
Estimated hours. Gut feeling about routine work is often off. Track for two weeks how long the task actually takes.
Costs that go away. When a process is set up anew, a tool sometimes becomes unnecessary. That saving belongs in the calculation, as does a transition period with overlapping subscriptions.
Checklist: measure honestly
- Track time spent on the task for two weeks, including interruptions
- Use fully loaded hourly cost, not just wages
- Estimate the handed-off share cautiously and measure it after launch
- Enter review and approval time as a new cost
- Note software that can be cancelled, with its notice period
- Repeat the same measurement after three months and compare
What cannot be counted in euros
Some effects do not show up in the formula but are real. Someone who no longer retypes quotes has time for customer conversations. A request answered the same day turns into an order more often. And because every step is logged, errors are easier to find than in a process that only exists in one person's head.
Do not convert these points into euros to make the number look better. Do note them as a second column in your assessment.
Explore practical examples, plan a supported migration, or compare your costs.
FAQ
What belongs in the hourly cost?
Besides gross pay, employer contributions, workplace, holidays and sick leave. If you are self-employed, use the amount you could otherwise bill in that time.
What if the agent takes over less than planned?
Then break-even moves further out. That is why it pays to estimate cautiously and measure after launch. Often the share can be increased by refining special cases.
How can I tell after three months whether it is worth it?
Repeat the time measurement from the start and compare routine time, review time and errors. If the net saving is higher than the monthly fee, you are on track.