Chain-of-Thought: Make the AI Reason Step by Step

Get consistent output by showing the AI examples instead of describing what you want.

Your prompt is probably the problem, not the model. Paste what you've been typing — it rewrites it into something that actually gets you what you want.

Paste a prompt that's not working. Get a sharper version and the reasons why.

Transform a rough prompt into a precise, high-performing one by systematically identifying ambiguity, missing constraints, weak instructions, and misaligned outputs, then correcting them through deliberate iteration.

Big tasks fail in one prompt. Chain them so each step builds on the last.

Stop rewriting the same prompt. Turn your best one into a fill-in-the-blanks template.

Generate responses that prioritise accuracy, evidence, and logical consistency over agreement, actively questioning assumptions, surfacing weaknesses, and stating uncertainty when claims cannot be supported.

Generate responses that prioritise independent analysis over agreement by evaluating claims on their merits, challenging assumptions, and presenting reasoned disagreement when warranted.

The one prompt that fixes half your bad AI answers. It asks you questions until it's 95% sure it gets what you want — then does it right the first time.

About to interview customers and worried you'll lead them or waste it? Practice on a realistic 'customer' first — then get coached on your questions and follow-ups.

Prep for the hard questions so the Q&A doesn't undo your whole talk.

Paste a blog post or article. Get 10 social posts pulled from it — different angles, different formats — so one piece of work feeds a whole week.