Idea: Generation of prompt for "chatbot" #1574
Replies: 2 comments 1 reply
|
I believe some users have experimented with C# scripts in Tabular Editor both to generate the prompt (i.e. output model metadata in a token-lean way based on the current selection), and to parse the results (for example, instructing the LLM to return the result as JSON or to write a C# script for model modifications). Tagging @data-goblin, @eugman and @vdvoorder in case they have some concrete experiences to share. That being said, for more advanced stuff, we recommend using the newly released |
|
OK, downloaded the CLI Public Preview. Question: Before investing time, is it anticipated that like AI Assistant it will be part of all license subscriptions? I only have the individual "base" plan (Power BI Desktop only). |
Uh oh!
There was an error while loading. Please reload this page.
New guy here, so please excuse if this is a dumb question. In looking through other comments, it looks like people are only using OpenAI and Anthropic models to date. Expen$ive!
Question: For those like myself who prefer to only have the $20/month subscription (ChatGPT Plus for me at the moment), have you ever considered a feature for AI Assistant, maybe called "Generate Prompt" that would provide a prompt that can be cut/pasted into a ChatGPT chat, resulting in output such as "add/update these calculations" or "modify your model as follows"?
I know I can feed a .bim file and DAX Script output to the chatbot and do it myself, but it is a very clumsy process. If using Tabular Editor 3 (I have the base version), there is no option to save the .bim file, so I need to use Tablular Editor 2. Also, using either Tabular Editor 3 SCRIPT DAX or C# in Tabular Editor 2, I need to generate some relevant DAX for any particular prompt. Then, I need to upload the files to ChatGPT, reference the files, yada-yada.
Anyway, just a "new guy" thought/question, while I'm here. Apologies' if it is a dumb idea. Just trying to stay within budget. Google, Anthropic and OpenAI all reduced token-usage a couple of months back, and I'm struggling to stay "in the loop" with AI.
All reactions