Skip to content
Chain Defenders
AI Systems

The Calculators

The AI Systems page has two calculators. They give a rough first number so you can decide whether a conversation is worth having. Your figures stay in your browser; we only count that a calculator was used.

What would automating a job return?

This estimates what a workflow costs you today, and what automating part of it would give back.

Inputs

  • Hours per week spent on the job.

  • Fully loaded hourly cost, meaning salary plus the cost of the seat, tools and management time.

  • The share of the job an automation could take over.

  • A one-off build cost.

Outputs

  • Hours given back per year.

  • Labour recovered per year.

  • Net benefit after the build cost.

  • Payback in months, shown on a chart of cumulative return over the first year.

How to read it

If the result says the job does not pay for itself, believe it. It usually means the hours figure covers a whole team rather than one person, or that the job is too small to be worth building. A large number is worth a proper look. The real figure comes from a measured baseline.

What does one finished AI task really cost?

This estimates what a finished task costs once you count every model call behind it.

Inputs

  • Model calls per finished task. Retries, checks, tool use and agent steps all raise it.

  • Tokens per call.

  • Price per million tokens.

  • Finished tasks per month.

Outputs

  • Monthly and yearly spend.

  • Cost per finished task, and per passing output.

  • A recoverable range per month, built from common levers such as caching, batching and routing, with vendor-stated figures labelled as such.

  • Spend at ten times the volume if nothing changes.

How to read it

If the number of calls per task is higher than you expected, that is the first place to look for waste. See Where Spend Leaks. Models cannot reliably predict their own token use, so enter figures you have measured, not guesses.

Important limits

  • Both calculators give estimates, not promises.

  • They depend on the numbers you enter. Rough inputs give a rough output.

  • They do not know your data, your quality needs or your provider's current pricing.

  • They are not quotes. Any saving we state in an engagement is measured against your own records.

Next step

If a result looks interesting, ask for a free first review. We will check the numbers against how your work really runs.