Spaces

How many hours could an AI agent give back to your central bank?

This calculator estimates the staff time an AI agent could give back to a monetary authority by handling routine visitor and publication questions. You enter your own message numbers and hourly staff cost.

Runs in your browser. Nothing you enter is sent anywhere or saved.

Before you start

At a central bank, only a part of the chat workload suits an agent. Anything about policy, markets, supervision, press or the law stays with people. What remains is practical: tour times, damaged notes, where a report is found and which desk to contact. This calculator looks only at that part.

Enter your own figures, and choose a cautious share for what an agent could handle. Be honest about the limits, since many messages at an institution like yours are sensitive. The result shows monthly hours spent, hours that could be given back, and what that time is worth in staff cost.

Every starting figure is an example. Change each one to your own.

Under your figures

Hours a month spent answering42
Hours an agent could give back, a month17
Worth in staff time, a month588

This is an estimate based on the figures you enter, so use your real numbers and a cautious share.

Reading your result

Treat the answer as a rough guide, not a target. Time given back is only useful if staff use it on something better, such as research support, visitor programmes or a faster press response. Decide in advance where the freed hours will go.

Be careful with the share. A figure of thirty to fifty percent may be reasonable for visitor and publication questions, but much lower if most of your messages are from the press or from supervised institutions. Start low and adjust after a trial.

  • Use a low share for the first estimate.
  • Count only routine, public questions.
  • Include the time staff spend reviewing the agent.
  • Decide where the freed time will be used.

FAQ

What share should we enter?

Begin with a cautious figure, perhaps a quarter to a half of visitor and publication questions, and far less overall. After a trial, use what you actually see in the chats to set a better number.

Does the result include the time to review the agent?

No. Reviewing chats takes staff time, especially early on. Subtract it from the hours given back to get a fairer picture, and expect the review time to fall as the notes improve.

Can I use this for a budget case?

Use it as one input, alongside your own judgement. It is an estimate from your figures and makes no promise of savings. Pair it with a trial so that you can replace guesses with real numbers.

More free tools for central banks and monetary authorities

Read the guide: A realistic role for an AI agent at a monetary authority.