The workbook guides you. Your chosen AI tool drafts. You check and keep the method.
Project requests arrive in separate messages. The client wants one checklist to keep the agreed scope clear. Next step: send a draft checklist for review.
You would like a single request checklist so the agreed scope is easier to find. As discussed, I will send a draft checklist for your review.
The need is accurate. The next step matches the call. No deadline or price has been invented.
Check, approve, then reuse.
Illustrative adaptation of the Sales Follow-Up worked example.
Build and edit your project in the workbook. Save your instructions, checks and fallback as a Runbook: your method for doing the task again.
Runbook excerpt
Start with familiar work and define the result you need.
Adapt the instructions. Try a normal case and an edge case.
Decide what is ready to use and save your reusable method.
Adapt one, or begin with a blank project and a task of your own.
Work from approved call notes to a draft that keeps the agreed next step intact.
Turn a table into a proposed summary, checking values and flagging gaps.
Turn process notes into instructions without inventing missing responsibilities.
Practical AI foundations, four reference guides and three fully worked, editable examples. Follow the workbook from a familiar task to tested instructions and a finished Runbook.
Already have access? Use your recovery code.
Not sure where to begin? Start with the free audit.
Research with clear limits
Published research has found measurable gains when generative AI was applied to specific work in controlled settings. The results make a carefully chosen task worth testing, not assuming.
15%more customer-support issues resolved per hour, on averagePublished study result
A field study followed 5,172 customer-support agents at one company using a generative AI assistant. Agents remained responsible for the conversation and could ignore or edit suggestions.
Context limit: Effects varied substantially. Less-experienced workers gained more, while the most experienced workers saw small gains in speed and small declines in quality.
Generative AI at Work, Erik Brynjolfsson, Danielle Li, Lindsey Raymond, The Quarterly Journal of Economics (2025).
Read the study in The Quarterly Journal of Economics40%less time on the assigned professional writing tasksPublished study result
A randomized experiment with 453 college-educated professionals found that average completion time fell by 40% and independently assessed output quality rose by 18% on short, occupation-specific writing tasks.
Context limit: The tasks were short and self-contained, without the company context or factual verification that much real business work requires.
Experimental evidence on the productivity effects of generative artificial intelligence, Shakked Noy, Whitney Zhang, Science (2023).
Read the study in ScienceThe practical question is not whether AI is universally productive. It is whether one bounded, recurring task has known inputs, a human reviewer, and a manual fallback, then earns an Adopt, Revise, or Stop decision through a small test.