- MUST READ: Submission requirements
- Entry form template
Team Horizon 2024
- Best use of AI
This is an attempt to design a solution that leverages Azure AI Translator service to automate document transcreations to different languages. The idea is to integrate XM Cloud with Batch document translations service, where by documents are pushed from Media Library in bulk into Azure Blob storage. They are then translated in bulk and are stored in a staged location within Azure Blob storage. The XM Cloud can then pull those back into Media Libary, for delivery to a global audience.
The vision is to enable businesses to take advantage of advances in AI natural language processing, and reduce costs and time it takes to transcreate business critical documents such as marketing material for their markets. Since Azure AI can translate documents whilst preserving their formats, this for me is the 'best use of AI' with XM Cloud to meet business needs.
I realised this was a huge undertaking, with one team member. However, I have started building the various componets to deliver the vision. This includes:
- An Azure AI Services Func app microservice (which islocated within the
DocuTranslatorsubfolder) - An SPE orechestration module built on XM Cloud to push documents from the Media Library to Azure Blob Storage for transcreation. Then pull the output and push it back into Media Library.
- An UX enhancement to XM Cloud Ribbon or Workflow enhancement to enable business users to incorporate this content automation in their processes
- I am planning to finalise the last two parts post the Hackathon due to timing constraints.
Below is a sample pdf image that has been transcreated with this service. The document structure is preserved, but not sure about the font.
⟹ Sorry no video this time round
⟹ Does your module rely on other Sitecore modules or frameworks?
- Azure AI Services Func app microservice, further details below
To run locally, refer to 'Development' section. Then Use the Sitecore CLI to serialize items into local environemnt
1. dotnet sitecore cloud login
2. dotnet sitecore connect --ref xmcloud --cm https://xmcloudcm.localhost --allow-write true -n default
3. dotnet sitecore ser pull -n "default"The Azure Func app solution requires the following settings to run locally. A sample local.settings.json has been included for reference
1. "docu_translation_endpoint": "Azure AI services document translation endpoint, more details in Usage instructions below"
2. "azure_key": "Azure AI services document translation key"
3. "docu_source_uri": "The URL for the source container containing documents to be translated"
4. "docu_target_uri": "The URL for the target container to which the translated documents are written"
5. "docu_target_lang": "The language code for the translated documents, e.g., fr for French"
6. "funcapp_api_key":"Your custom function app API key"To invoke the Func app locally, create POST request using a tool like Postman, as shown below. Use X-Api-Key to pass required funcapp_api_keyon the header
The published Func app on Azure portal is configured on /sitecore/system/Modules/PowerShell/Script Library/teamhorizon2024/Azure API Settings.
As you can see, here we configure both the Func app URL and API key on this item.
Document Translation is only supported in the S1 Standard Service Plan (Pay-as-you-go) or in the D3 Volume Discount Plan. This service needs to be provisioned on Azure Portal Azure AI Translator, if you are running your own local instance of the Func app. For the published Func app on Azure portal, you do not need to configure Document Translation.
⟹ Adopted from XM Cloud Starter Kit (Next JS)
-
In an ADMIN terminal:
.\init.ps1 -InitEnv -LicenseXmlPath "C:\path\to\license.xml" -AdminPassword "DesiredAdminPassword"
-
Restart your terminal and run:
.\up.ps1