Conversation
jdbaldry
left a comment
There was a problem hiding this comment.
Very cool! Will let Chris take the lead in reviewing the guidance but this looks super useful
GrafanaWriter
left a comment
There was a problem hiding this comment.
❤️ this addition to our Toolkit, @grafsean. Thank you!
One q: is there any where we can explicitly tell the AI tool to not make inferences?
|
@GrafanaWriter I've left instructions of that nature out for the prompter to control and focussed on the structure and content of learning journeys. What I find works really well is saying, use a this resource for that section. I think there are smarter workflows we can introduce which includes testable docs, that will catch incorrect inferences. Software engineering has these more complete experiences, if an LLM writes code, it can verify it. We don't quite have that yet. |
chri2547
left a comment
There was a problem hiding this comment.
This is fabulous. Can't wait to use it!
@grafsean are usage guidelines like this documented anywhere? I'm interested in using this but not sure where to put these files in the website repo and any tips or techniques you have would be very useful. |
Adds instructions for learning journeys.
The instructions work best for getting you started with a content structure you can iterate on.
I've experimented creating a handful of journeys using existing docs and blog posts as context.
It works best when specifying the specific steps and content you'd like, for example, "Create a learning journey for App O11y, use the attached resources as context. Use the following steps: 1. set up the Grafana Java SDK, 2 set up Alloy, 3 send to Cloud, 4 end journey. Use article for the Alloy setup."
I've found Claude Sonnet gives better results over Gemini and OpenAI.