Companion videos

Using ChatGPT to write emails you actually want to send - companion videos

The article uses approved, minimal context and a facts card to draft an email, then checks every name, date and commitment before sending. The video offers optional prompting ideas. A model's critique can suggest edits; it cannot verify facts or predict the recipient's reaction.

Primary pick

10:49
4 ChatGPT Hacks that Cut My Workload in Half

Jeff Su

The creator lists a "Red Team Technique" chapter at 06:06 about asking AI to critique its output. Try a bounded critique of tone or clarity on fictional or explicitly approved facts. Treat proposed reactions as hypotheses, not evidence about a real recipient.

What you should get from this: Request quoted suggestions on tone or clarity, then verify the draft against the approved facts and intended meaning.

Watch or know first: A fictional or deliberately non-sensitive exercise. Use real correspondence only when the exact tool, account, feature and purpose are approved for its data.

AI Expert note: Use the critique chapter as an optional example. The title's workload saving is the creator's claim, not a measured AIExpert result or a promise for your workflow. Keep private threads out, verify every proposed change and check recipients and attachments in the email app.

Open video page