Claude
9 hours 40 minutes, honestly counted
Lilach Bullock runs a solo marketing and consulting business. For 30 days in September 2026 she logged every task where AI helped, in a sheet with five columns. The log counted draft time and every minute she spent editing and fixing the output. Claude drafted client proposals and long first drafts. ChatGPT drafted warm-lead replies and invoice-chasing notes. Otter.ai transcribed calls into follow-up emails. Descript captioned short videos. Subscriptions cost 71 pounds for the month.
How it wentThe net saving was 9 hours 40 minutes across seven logged tasks, about 7.30 pounds per hour saved. Proposals fell from 45 minutes to about 22 minutes each. Blog first drafts nearly broke even, with a 55-minute rewrite after a 4-minute draft. Video captions lost about 40 minutes across five of nine videos. Her rule: AI saves time on structured, repeatable, low-personality tasks, and costs time on voice-driven work.
Try it · based on the source's workflow
Keep a time log for [number] days. For every task where you help me, record the task name, my old time for it, your draft time, and my edit and fix time. Use this table: [task, old minutes, draft minutes, edit minutes]. At the end, total the true time saved per task and flag any task where the total is worse than my old way.
A starting template based on the workflow the source describes. Review it before you use it; you are responsible for what it does.
9 hours 40 minutes, honestly counted — from The SmallOffice Brief, a daily brief of real AI cases for small businesses: https://smalloffice.space/case-9-hours-40-minutes-honestly-counted