# Companion plugin for Obsidian
Autocomplete using AI, including ChatGPT, and local model (e.g., via Ollama).
Overview:
## Pre-requisites
If you want to use [[Ollama]] with this plugin, you first need to install it on your machine.
## Configuration
Settings recommended by Matt Williams (cfr references below):
- Enable "Streaming mode"
- Set delay to 500ms
- Accept: One word at a time
- Minimum display length: 10
- Retrigger threshold: 10
- Provider: Ollama
- API route: http://localhost:11434
- Model: gemma2:9b (or any other model you want)
- Temperature: .2
User prompt:
```
{{#context}}Context:\n\n{{context}}\n\n=================================\n{{/context}}
The following text has been written by the user. You will continue writing the next few words of the text as if you were the original writer. Do not begin the text with '...' and don't summarize the text.
{{last_line}}
```
## References
- Plugin sources: https://github.com/rizerphe/obsidian-companion
Great introduction tutorial from one of the original [[Ollama]] maintainers:
## Incoming links
<!-- QueryToSerialize: LIST FROM [[Companion plugin for Obsidian]] WHERE public_note = true SORT file.name ASC -->
<!-- SerializedQuery: LIST FROM [[Companion plugin for Obsidian]] WHERE public_note = true SORT file.name ASC -->
- [[33.02 Content]]
- [[AI (MoC)]]
- [[Obsidian (MoC)]]
- [[Ollama]]
- [[Useful resources for Obsidian]]
<!-- SerializedQuery END -->