What's your AI really worth?
Nobody can say what the AI budget produced
The numbers come from runs that actually happened, not from a projection of what might. Confirmed and estimated stay separate, so a headline number is only the part somebody checked.
Powered By
The Challenge
- The board asks what the AI budget produced, and the answer is a story about how much everyone likes it
- Every vendor hands you a projection. You need something that already happened
- You can count seats and you can count tokens. Neither of those is a result
- The number somebody produces in a spreadsheet is a guess wearing a confident face
- Renewal is in six weeks and there is nothing you would put in front of finance
The Runwork Solution
- Each number comes from runs that happened, with the record of those runs behind it
- Confirmed and estimated are kept apart and never added together, so a headline number is only the part somebody checked
- A run is credited with the manual work it replaced: an hourly job that replaced a weekly chore is credited weekly, not 168 times
- What you pay for Runwork sits beside the savings with the subtraction shown, and AI subscriptions billed elsewhere are excluded and said to be
- Five reports written to be forwarded, produced the same way every time, with no AI anywhere in the calculation
What your team gets
A number with a name against it
Every savings figure records who checked it. Confirmed and estimated sit in separate columns and are never added into a single headline, because only one of them would still be standing after a question.
The cap that makes it believable
A job that runs every hour but replaced a weekly chore is credited weekly. Crediting it 168 times would produce a much better number and a much shorter meeting.
Our own cost, subtracted in front of you
What you pay us sits beside what the work saved, with the subtraction shown. AI subscriptions billed somewhere else are left out, and the report says they are left out.
Work that ran while nobody was there
Runs outside working hours, runs while the owner was away, and the longest stretch without a gap. That is capacity that did not exist before.
A report you can forward
Five reports for five audiences, produced the same way every time. There is no AI in the calculation, so the same inputs give the same output whoever runs it.
Where we decline to give you a number
Hours spent rebuilding something are never turned into money, and work that never existed before is shown as new capacity rather than as savings. Both would be easy to dollarize and neither would be honest.
A Day in the Life
9:00 AM. The email from finance. They want to know what the AI line item bought this year, before renewal. It is a fair question and until now the honest answer has been that people seem happier.
10:00 AM. You open the confirmed column. The savings report keeps confirmed and estimated apart and does not add them together. You take the confirmed figure, which is smaller than the one you would like, and it is the one that will still be standing after somebody asks where it came from.
11:00 AM. The question you already have an answer to. Somebody will ask whether an hourly job really saved an hour every hour. It did not, and the report never said it did: a run is credited with the manual work it replaced, so an hourly job that replaced a weekly chore is credited weekly.
1:00 PM. Your own invoice, on the same page. What you pay us is set against what the work saved, with the subtraction shown rather than implied. The report also volunteers what is not in it, including AI subscriptions billed somewhere else.
3:00 PM. The part nobody expected. Work that ran outside working hours, and work that ran while the person who built it was on leave. Nobody was at a keyboard for any of it, which makes it capacity rather than effort moved around.
4:30 PM. You forward it. The report is built to be sent to somebody else, so you attach it rather than rebuilding it as slides. Finance reads the same numbers you read, including the ones we declined to inflate.
Frequently Asked Questions
Where do these numbers come from?
How is this different from a value assessment?
What stops these numbers from being flattering?
Can I compare our results against other companies?
Does AI produce these reports?
Related Use Cases
Start keeping what your team figures out.
Start a 14-day free trial. No credit card required.