The Most Useful Plugin For Obsidian| GPT-3 AI Writer

A free way to use GPT-3 writing help inside Obsidian hinges on one practical setup: install the “Text Generator” plugin, add an OpenAI API key, and then generate drafts directly from the note context. Instead of paying for standalone AI writing services like Jasper or Copy.ai, the workflow routes requests through OpenAI’s API—so costs scale with usage and the tool can be used as a daily writing and knowledge-management assistant.

The setup starts in Obsidian’s Settings, where the “Text Generator” plugin is installed and enabled. The plugin then asks for an API key, which is obtained from OpenAI’s account dashboard under API keys. After the key is added, the plugin’s options map to OpenAI’s model and generation controls. Model selection determines both speed and power: the “engine” setting uses either the more capable DaVinci model or the faster Ada model. The transcript also breaks down pricing in token terms—roughly 1,000 tokens are about 750 words—showing that Ada is far cheaper per token than DaVinci. OpenAI’s free tier is positioned as the on-ramp: 18 in free credit for the first three months, with the presenter noting only a small amount consumed after a couple of days.

Within Obsidian, the plugin’s generation behavior is tuned using parameters like Max tokens (the maximum output length per generation), temperature (randomness/creativity), and frequency penalty (reducing repeated words). There’s also an optional status bar indicator to show when generation is running. Hotkeys provide quick control: one set adjusts Max tokens temporarily, while others trigger generation with or without metadata context.

The core usage concept is “context.” The AI can generate from whatever the cursor is currently on—either a line, the whole node if the cursor sits on an empty line, or a selected text span. For more targeted results, the plugin can use metadata from the note (such as a title and keywords) as the prompt context. Two generation commands are demonstrated: generating text normally (e.g., using Ctrl J) and generating text using metadata (e.g., Ctrl Alt J). The examples show the plugin drafting content about topics like reading habits, then continuing the draft with another generation call.

Beyond drafting paragraphs, the same mechanism supports idea generation and brainstorming. Users can request multiple ideas for an article topic (like reading more books) or for a niche YouTube channel, then iterate by generating additional text. The transcript also mentions using the tool for summaries and conclusions, though one attempt doesn’t work in the presenter’s setup.

Overall, the takeaway is a workflow: generate drafts, expand notes, and brainstorm ideas inside Obsidian by combining OpenAI’s API with the Text Generator plugin—keeping costs low through free credit and token-based pricing while maintaining a note-first writing process.