Simulator
What does your repetitive task cost when AI does it?
Pick a task, then enter how often it runs and how long it takes. With a frontier AI, every run uses tokens and the bill climbs. With a Nestor agent, a run calls no model: 0 tokens.
Over 12 months
Frontier AI
€1,347
over a year
Nestor agent
€0
per run: the program agent calls no model
Frontier AI monthly bill: month 1 to month 12 : €63 → €180
- 756
- runs per year
- 134.7M
- tokens used by the frontier AI
- 252
- hours of work involved
Illustration. Figures are not contractual: the assumptions above are orders of magnitude, to be replaced with yours.
How it's calculated
- Runs per month = frequency × people (21 working days per month, 4.33 weeks per month).
- Tokens per run = task duration × tokens per minute: the frontier AI rereads your documents and writes every time.
- Bill = tokens × price of one million tokens, with usage growing from one month to the next.
- Nestor: a program agent, written once by AI then tested on your files, runs with no model: 0 tokens per run. Building it, then each change, uses tokens once: that cost is estimated and shown in Nestor, and isn't included here.
- Default price: public rates of Claude Opus 5 (5 input, 25 output per million tokens, dollars taken as euros), blended at 3 parts input to 1 output, i.e. 10. Editable.
And on your real tasks?
In an audit, we start from your volumes and files to price what can become an agent.