Free tools · AI Time‑Back Calculator
How many hours could AI give your team back?
Put in roughly how many hours a week your team spends on seven common jobs. You’ll see the hours AI could take back, based on published research, and the three jobs to start with.
Hours a week, across your whole team
Free · keep this result
Email me this report as a PDF.
Your result above, written up properly and sent from Keith’s own address, in the next few minutes.
We’ll email your report and may follow up on WhatsApp about it, never with spam.
Free · written by hand
Want the full picture of your marketing?
A calculator only knows the hours you give it. The full analysis is written by hand for your business, free, within 1 working day:
- Which AI tools name you when customers ask, and who they name instead
- Your website page by page: is each one doing its job?
- Your Google profile next to your nearest rivals
- What to fix first, in order, and what each fix wins
Free and no obligation. We’ll reply by email or WhatsApp, and never spam you.
Where the numbers come fromEvery range, and the study behind it.
Each range is the share of that job’s hours AI could take back. The low end is close to AI used by hand, and the high end to AI built into the job with a person checking.
| Job | Time back | Based on |
|---|---|---|
| Writing posts, emails and articles | 15% to 35% | Noy and Zhang, Humlum and Vestergaard |
| Answering customer enquiries | 10% to 20% | Brynjolfsson, Li and Raymond |
| Quotes, proposals and documents | 10% to 25% | Noy and ZhangThere’s no study of quotes on their own, so this uses the writing research, set lower. |
| Reports, summaries and meeting notes | 15% to 30% | Dell’Acqua, Mollick and others |
| Sales follow‑ups and outreach | 15% to 30% | Noy and ZhangThere’s no study of sales outreach on its own, so this uses the writing research. |
| Admin, data entry and scheduling | 10% to 25% | Bick, Blandin and DemingThere’s no controlled study yet, so this sits a little above the St. Louis Fed average. |
| Spreadsheets and number work | 15% to 30% | Cui, Demirer and others |
- Noy and Zhang (2023), in ScienceIn a controlled experiment with 453 professionals, ChatGPT cut the time spent on writing tasks by 40%.Read the study
- Brynjolfsson, Li and Raymond (2023), NBERCustomer support agents given an AI assistant resolved 14% more issues an hour on average, and newer agents 34% more.Read the study
- Dell’Acqua, Mollick and others (2023), Harvard Business School and BCGConsultants using GPT‑4 finished their tasks 25% faster and completed 12% more of them.Read the study
- Cui, Demirer and others (2024), field experiments at three companiesAcross 4,867 developers, those given an AI assistant completed 26% more tasks.Read the study
- Bick, Blandin and Deming (2025), Federal Reserve Bank of St. LouisPeople who used generative AI at work saved 5.4% of their work hours on average.Read the study
- Humlum and Vestergaard (2025), NBERAcross 25,000 workers in Denmark, people using AI chatbots reported saving about 3% of their hours.Read the study
Questions
About the calculator.
How does the AI Time-Back Calculator work it out?
For each job, it multiplies your weekly hours by a low and a high share of time saved, taken from published studies.
The low end is close to what people report when they use AI by hand. The high end is closer to what controlled studies found when AI is part of the job, with a person checking the work.
Why isn’t video editing in the calculator?
No independent study has measured how much time AI saves on video editing yet, only claims from companies selling the tools. We edit video with code ourselves, and we’d rather show you that on a call than put an unproven number here.
Why are the numbers lower than other AI calculators?
Many calculators use the best result from a lab study. Surveys of real workplaces find smaller savings, around 3 to 5% of all work hours for people using chatbots by hand.
So the ranges here lean low, and the hours really show up once AI is built into the job.
Are my answers saved?
The calculator runs in your browser, and we don’t see anything you type. If you ask for the PDF or the free analysis, we keep what you type in that form, because that’s how we deliver it. Like the rest of the site, the page counts visits with Google Analytics and Microsoft Clarity, as the privacy policy explains.