docs.neptuneplayer.com
Warning: If this is a folder, all contents including subfolders and documents will be deleted.
Define path-based access rules for sections of your wiki, then assign users to groups in the Users tab. Rules are evaluated in order. First match wins.
Create and manage backups of your wiki data. Backups include all documents, images, and configuration files.
When a query needs actual understanding ("farm boy learns magic and fights his dad"), Neptune hands it to a language model. AI Search is built into the same search bar as everything else: no mode to switch, no separate screen.
The easiest provider is Neptune AI, the zero-setup hosted service included with Neptune Pro. You can also use a supported bring-your-own provider.
You can also trigger it directly: when a search comes up empty, an Ask AI chip appears under the results.
| Query | Finds |
|---|---|
| "farm boy learns magic fights dad" | Star Wars |
| "movie where they shrink the scientists" | Fantastic Voyage, Innerspace |
| "cooking rat movie" | Ratatouille |
| "detective can't remember anything" | Memento |
For finding a specific episode of a show, use Episode Finder on the show's page instead.
AI Search needs a language model. Configure one in Settings > Deep Learning:
| Provider | Type | Notes |
|---|---|---|
| Neptune AI | Hosted | Built in with Neptune Pro; no keys to manage |
| Ollama | Local | Runs on your own hardware, fully private |
| Gemini | Cloud | Bring your own API key |
| OpenRouter | Cloud | One key, many models |
Claude, OpenAI, and GLM appear in the provider roadmap but are not active providers yet.
Add several providers and use Set Model Priority to control the fallback order. If your primary is unreachable, the next one takes over. The same providers power Episode Finder, Title Insights, Person Insights, and other AI features.
AI queries take a few seconds, depending on the provider and model. If the model times out or fails, you still get the instant results; AI Search only ever adds.
| Limitation | Why |
|---|---|
| Home videos and niche content | The model can only reason about titles it knows |
| Very new releases | May post-date the model's knowledge |
| Non-media queries | It searches your library, it isn't a chatbot |
Only the query text and basic result context are sent to the provider, and only when AI Search actually runs. Nothing is uploaded during normal instant searches. With a local provider like Ollama, nothing leaves your network at all.