Prompting LLMs is like going down on a woman

- Always trying new techniques, never quite sure if you're doing it right.
- When you finally find the right prompt, you stick with it like you found the magic rhythm. 
- Each AI has its quirks: Claude gets shy if you're too aggressive, Llama will try anything once. 
- Sometimes they're into something one day and completely against it the next, with zero explanation
- Both will occasionally give you a completely unexpected response that makes you question everything you thought you knew
- Just like with "I have a headache," there's no arguing with "I cannot assist with that task"
- Context windows are like stamina - you hit the limit at the worst possible moment
- Rate limits are basically the AI's version of "Not tonight, I'm tired"
- The more specific you are about what you want, the better the results
- If you keep asking 'Was that good?' after every response, you're definitely doing it wrong.

Ex-VC. I like making bread (literally and figuratively) 