# Diagrams

TRS implements basic indexing and retrieved-augmented generation pipelines. 

<figure><img src="https://976894563-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FYZqoOK6qjDt28AZLxbEt%2Fuploads%2FbvVeQQ7nlN0w7tDrB81H%2Findexing.png?alt=media&#x26;token=3051ace9-f984-4559-9584-d5477c41424c" alt=""><figcaption><p>Data indexing pipeline</p></figcaption></figure>

<figure><img src="https://976894563-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FYZqoOK6qjDt28AZLxbEt%2Fuploads%2FTPf0ZuQMGz6HLz3j4cHe%2Fretrieval.png?alt=media&#x26;token=a0b9565d-c3a1-4cf8-9f7c-cd3cbbd4ba4b" alt=""><figcaption><p>Retrieval pipeline</p></figcaption></figure>

#### Planning

Planned upgrades to the pipelines include:

* Addition of summary index
  * Summarized text is stored in it's own index
  * Retrieval can query both the document and summary index then synthesize answer across both sources
* Better management of previously processed URLs