← What to do

Claude 101: A free guide to everyday Claude habits from Ruben Hassid

This Substack post walks through 17 habits for getting better results from Claude, Anthropic's AI assistant. Several of them are beginner-friendly. When Claude gives a bad answer, you can edit your earlier message instead of arguing with it. You can keep one Project per client with account memory turned off so details don't mix between clients. You can tell Claude what 'done' looks like. You can also export a bank CSV, a simple spreadsheet file, instead of connecting your account. The author says keeping memory on makes Claude '28% better,' citing Anthropic research. Treat that figure as his summary of the research, not a guarantee.

For
People who already use Claude a little and want practical habits, including anyone who runs a business and juggles several clients in one AI account. No technical experience is needed for most tips, but the GitHub tips are more advanced.
Cost
The 17 tips shown are free to read. The post says 10 more tips are available only to paid subscribers, and the subscription price is not listed in the post. Claude's own plan costs are not covered.
Access
Open to anyone online. You need a Claude account to try the tips. The post does not say which Claude plans include features like memory, Projects, or the Google Docs extension.
Byte's take

The free section is worth a read if you already use Claude and keep getting messy or repetitive answers, and the per-client Project tip is especially useful if you run a business. Skip the paywall unless you want more, and treat the author's percentages and privacy claims as his reading, then check them against Claude's own settings.

Your moveOpen the post and try two tips in Claude today. First, edit your prompt after a bad answer instead of replying 'that's wrong.' Second, create a Project for one client and turn off 'Use account memory' inside it.

Start here

More in What to do