Series

One Useful Task

AI at Work: One Useful Task at a Time

Practical Work Recipes for people who already opened the chat window. One real job, one method, one output you can check. Task → method → output → verify → limits. No coding required for this track.

You’ve tried ChatGPT or Claude at work. You don’t need a chatbot wrapper — you need a habit you can repeat on a busy Tuesday.

01
task
02
method
03
output
04
verify
05
limits

Read now

01Work Recipe

Vague Friday Requests Aren’t Briefs — Clarify First

Turn a vague Friday ask into a draft management discussion brief: outcomes, decisions, next steps, and agenda — before you write the deliverable.

4 min · Turn a vague request into a draft work brief a human can confirm.

02Work Recipe

Sounds Good ≠ Correct — Own the Output

Confidence is a style. Correctness is ownership. Catch confident-but-wrong output before it leaves your desk — the tool changed; the bar didn’t.

5 min · Catch confident-but-wrong output before it leaves your desk — same ownership as any other draft.

03Work Recipe

Before You Paste — What Shouldn’t Go Into the Chat

Removing people’s names is not the same as making information safe to share. Match the tool tier to what you’re allowed to paste — ZDR doesn’t skip judgment.

5 min · Prepare a safer practice version of a work document or decide what to paste into which tool.

Later in the map

Outlines only for now. They ship as Work Recipes on this site first, then Medium, then Substack.

04Coming next

Brief the Model Like a Colleague

Give role, context, and a definition of done — the way you’d brief a sharp new teammate.