> ## Documentation Index
> Fetch the complete documentation index at: https://docs.clipsx.app/llms.txt
> Use this file to discover all available pages before exploring further.

# Local AI with Ollama

> Connect a loopback Ollama service for optional Meaning Search and Recall features.

ClipsX can use an Ollama service running on your device. It does not install
Ollama, download models automatically, or silently send clipboard content to a
hosted model provider.

<Steps>
  <Step title="Install and start Ollama">
    Follow the [official Ollama installation instructions](https://ollama.com/download),
    then start its local service. The default endpoint is
    `http://localhost:11434`.
  </Step>

  <Step title="Connect ClipsX">
    In ClipsX, open **Intelligence**, choose **Models**, enter the loopback
    endpoint, and select **Connect**. ClipsX lists models after the endpoint
    responds successfully.
  </Step>

  <Step title="Enable Meaning Search">
    Select an installed model that Ollama reports as supporting text embeddings.
    Indexing runs in the background, while exact text search remains available.
  </Step>

  <Step title="Enable Recall when needed">
    Under **Local Text Generation**, choose a model that reports generation
    support. This can be different from the embedding model.
  </Step>
</Steps>

## Recall safety model

Recall answers questions from a visible, bounded selection of history. Inspect
its numbered citations before relying on an answer. Detected secret clips are
excluded unless you override that protection, and generated answers can be
wrong.

<Note>
  Prompts and answers do not become canonical clip metadata. Temporary Recall
  conversations expire after inactivity.
</Note>
