When I made the switch from ChatGPT to Claude, here’s what I wish I knew..
1. Know when to use a Chat vs a Project
If a conversation is going to run longer than 5 to 10 messages, make it a Project. Projects carry consistent context and instructions every single time and can keep reference documents. Think of them like a mini preferences layer built around your specific goal or “project” 😉
Chats are for quick, contained tasks.
2. Set up Memory and Preferences
Don’t try to write perfect preferences on day one. Use Claude for a while, notice what consistently bugs you, and add a rule to eliminate it.
3. Be specific upfront
Tell Claude what you want before it starts. Format, tone, length, all of it. It saves you correction rounds and keeps the thread clean.
4. Edit, don’t follow up
When Claude misses the mark, edit your original message and regenerate instead of sending a follow-up. Keeps your thread shorter, your context cleaner, and it saves tokens. Win-win.
5. Ask Claude to write you a fresh start prompt
When a thread feels heavy, ask Claude to summarize the conversation into a prompt you can drop into a new chat or Project. And if you’re doing this a lot, see point 1.
6. Bonus tip: double check the work
Claude is a powerful tool. It is not infallible. Always review before you ship.
Each of these deserves its own post.. More coming. ☺️