> For the complete documentation index, see [llms.txt](https://llmvision.gitbook.io/getting-started/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://llmvision.gitbook.io/getting-started/introduction.md).

# Introduction

<figure><img src="/files/0unjkR0POzs69krPnC6e" alt="" width="563"><figcaption><p>Timeline Card</p></figcaption></figure>

## Introduction

### What is LLM Vision?

LLM Vision is a Home Assistant integration that leverages multimodal LLMs to analyze images, videos, and live camera feeds, answer questions about them, and update sensors based on extracted data. \
The integration can optionally store analyzed images along with their AI summaries as events. These events can be viewed on the dashboard using the [LLM Vision Timeline Card](/getting-started/setup/timeline-card.md).

The integration also has a built-in memory to store reference images along with a description to easily [recognize people, pets or cars](#user-content-fn-1)[^1].

### Architecture

The diagram below is a high-level overview of the different components.

<figure><img src="/files/1JfKMSMwWMFlb2NUhdKH" alt=""><figcaption><p>High-level overview of LLM Vision components</p></figcaption></figure>

LLM Vision consists of multiple key parts, which need to be installed separately:

* Integration (LLM Vision)
* Blueprint (`event_summary`)
* Timeline Card

At its core, the **integration** handles processing camera streams and sending them to an AI provider. \
It consists of a Settings config entry, as well as config entries for each AI provider such as OpenAI, Anthropic, etc.

**The blueprint** is a ready-made automation that is easy to use but customizable. It is triggered by a camera or binary sensor state change. When triggered, `stream_analyzer` is called, which returns a summary of what is happening in the camera feed. This summary is then sent to your phone via Home Assistant app notification. Optionally, the analyzed "event", can also be saved in the timeline.

**Timeline** is configured in **Settings** and adds the ability to store events so they can be viewed on the dashboard through the [**Timeline Card**](/getting-started/setup/timeline-card.md).

Continue, to learn how to install and set up these components.

[^1]: LLM Vision uses large language models. Do not use LLM Vision to automatically unlock doors or for similarly critical actions!&#x20;