Okay, this isn’t some groundbreaking thing. This has been something you can do in Obsidian for quite some time. I decided to share this workflow because I came back to using genAI in Obsidian after taking some time off.
Backstory
I had a test two days ago. 3 days ago, I knew relatively nothing about the course. “It’d take me way more time to devour the course materials” I thought to myself, and I was right. I’m not someone who simply memorizes as much formulas as their brain decides not to OD on—I hate that. I prefer understanding concepts to memorizing*. Don’t get me wrong, as an engineering student it’s necessary to memorize. I’ve memorized formulas I did not understand—that’s part of the job description. LMAO.
*(Tell me you’re really smart without telling me you’re equally not so smart.)
- The problem with writing this blog post is that I need to take screenshots of Google studios so I can show you where the default generated API key is.
- Then we need to be able to breakdown the options in Text Generator, so I can easily show you what options I had enabled to make this thing work. I think this is better with OpenAI and Claude API keys.
- Then I need to show you an example of a generated file output with the agent I might create, say Rosé or some other name (I could check names of those Microsoft voice models—Aria seems like a good one).
- Other generation types would be just finding the answer to something as you have it by default in Obsidian, by using the
Ctrl+Shift+Ghotkey to generate some text based on the last line (which is ideally the prompt). - Finally, you have preferences of whether to use Gemini or not, Claude is best for the Rosé archetype I have built. DeepSeek might be a game changer due to it’s open nature and the very affordable pricing it’s starting with.
What do you think?