# December 22: Support for Dropbox, Box, Intercom and Zendesk feeds, OpenAI o1, Gemini 2.0, bug fixes ### New Features * :bulb: Graphlit now supports Dropbox feeds for ingesting files on the Dropbox cloud service. Dropbox feeds require your `appKey`, `appSecret`, `redirectUri`and `refreshToken`to be assigned. The feed also accepts an optional `path`parameter to read files from a specific Dropbox folder. * :bulb: Graphlit now supports Box feeds for ingesting files on the Box cloud service. Box feeds require your `clientId`, `clientSecret`, `redirectUri`and `refreshToken`to be assigned. * :bulb: Graphlit now supports Intercom feeds for ingesting Intercom Articles and Tickets. We will ingest Intercom Articles as `PAGE`content type, and Tickets as `ISSUE`content type. Intercom feeds require the `accessToken`property to be assigned. * :bulb: Graphlit now supports Zendesk feeds for ingesting Zendesk Articles and Tickets. We will ingest Zendesk Articles as `PAGE`content type, and Tickets as `ISSUE`content type. Zendesk feeds require the `accessToken`property and your Zendesk subdomain to be assigned. * Graphlit now supports the latest OpenAI o1 model, with the model enums `O1_200k`and `O1_200k_20241217`. * Graphlit now supports the latest Gemini Flash 2.0 Experimental model, with the model enum `GEMINI_2_0_FLASH_EXPERIMENTAL`. * Graphlit now supports the latest Cohere R7B model, with the model enum `COMMAND_R7B_202412`. * Graphlit now supports returning the low-level details from prompting RAG conversations, by adding the `includeDetails`parameter and setting to True. This includes details on the number of sources, the exact list of messages provided to the LLM, and more. * We have added support for filtering of observables, such as Person or Organization, by URI property. * We have added the ability to bypass semantic search in content retrieval with conversations. You can assign `NONE`for the conversation search type, and it will ignore the user prompt when retrieving content. It will inject all contents resulting from the content filter into the RAG prompt context. * We have added a new `createdInLast`property to all entity filters, which allows easier filtering of entities created within a recent time period. Also, we have added a new `inLast`property to the content filter, which allows easier filtering of content authored within a recent time period. For example, find all images taken in the last 3 days, or find me all emails I received yesterday. * We have added support for the latest Azure AI Document Intelligence models, with enums `US_PAY_STUB`, `US_BANK_STATEMENT`, and `US_BANK_CHECK.` * We have added support for Google Drive and OneDrive feeds to ingest specific files by providing a list of file identifiers (`files`), in addition to the folder identifier (`folderId`). If files identifiers are provided, they take precedence over the folder identifier. * :zap: For projects upgraded to the Starter Tier after Dec 9, 2024, we have removed the content items limit. Now you can store an unlimited number of content items (i.e. files, web pages, Slack messages) on the Starter or Growth Tiers. If you have an existing project on the Starter Tier, please reach out and we will manually remove that content item limit on the project. ### Bugs Fixed * GPLA-3529: Can't assign collection to multitenant content * GPLA-3579: Should decode HTML characters when parsing HTML email * GPLA-3576: Ingesting content in-place doesn't handle isSynchronous properly * GPLA-3457: IsFeedDone doesn't return True for finished feed with no contents * GPLA-3572: Not handling HTTP 400 error on uploading from URI --- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://changelog.graphlit.dev/december-2024/december-22-support-for-dropbox-box-intercom-and-zendesk-feeds-openai-o1-gemini-2.0-bug-fixes.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.