I’ve also noticed that one of the biggest problems that LLMs solve for me is reducing the friction to get started.
For me, I found that getting a long response can introduce new friction, especially at the end of a long day. What works for me is to structure the prompt to prepare the full response, but present it one step at a time. It encourages some back and forth in each area before moving to the next section.
This has two benefits:
1. I can always hit pause and return to the conversation tomorrow. I get to capture my thinking mid-way through the process.
2. I get to nudge and course correct the AI. My prompts follow a similar structure to what you presented, which benefits from top-down fixes.
This is a good idea, thank you! I love another aspect of this technique: by capturing my worries, it frees my mind to do thinking and rest assured that these are not forgotten by quietly marinating in my brain until I have time to work with them.
One risk of using LLMs to prioritize these is that they will pull towards the average. (Though this can be a benefit too.) If you need radically new ideas and creative approaches, it might be counterproductive.
I’ve also noticed that one of the biggest problems that LLMs solve for me is reducing the friction to get started.
For me, I found that getting a long response can introduce new friction, especially at the end of a long day. What works for me is to structure the prompt to prepare the full response, but present it one step at a time. It encourages some back and forth in each area before moving to the next section.
This has two benefits:
1. I can always hit pause and return to the conversation tomorrow. I get to capture my thinking mid-way through the process.
2. I get to nudge and course correct the AI. My prompts follow a similar structure to what you presented, which benefits from top-down fixes.
This is a good idea, thank you! I love another aspect of this technique: by capturing my worries, it frees my mind to do thinking and rest assured that these are not forgotten by quietly marinating in my brain until I have time to work with them.
One risk of using LLMs to prioritize these is that they will pull towards the average. (Though this can be a benefit too.) If you need radically new ideas and creative approaches, it might be counterproductive.
Loved this James. Great to read something practical for using LLM's for engineering management - please keep writing more about this topic!