indexer is failing with 403 error. 'Could not execute skill because the Web Api request failed.'

Hi @Shilpi Verma

Thank you for reaching us regarding the issue.

We have begun looking into it and will share our suggestions with you as soon as possible. If you have any additional details or concerns in the meantime, please let us know. We appreciate your patience while we work on this.

Hi, do we have any updates on it? I cannot raise a support ticket so is there any alternative to connect with technical support via Microsoft teams meeting. Please help, I am stuck and not able to proceed with any AI search related tutorial.

Hi @Shilpi Verma

Sorry for the late reply.

Thank you for the detailed description. The 403 error and the missing models are both expected behaviors when working in a fully private environment with the Import data (vectorize) wizard.

1. Indexer 403 error: “Public access is disabled. Please configure private endpoint.”

The Import data wizard and the indexer it creates cannot reach resources that have public network access disabled unless you configure shared private links and force the indexer to run in the private execution environment.

Required steps (from Microsoft docs):

1. On your Azure AI Search service go to Networking > Shared private access > + Add and create shared private links for:
   • Storage account (Microsoft.Storage/storageAccounts, group ID blob or dfs)
   • Your Foundry / Cognitive Services / Azure OpenAI resource (Microsoft.CognitiveServices/accounts, appropriate group ID such as openai_account, cognitiveservices_account, or foundry_account)
2. Approve the private endpoint connections on each target resource (Storage account and Foundry resource) under Networking > Private endpoint connections.
3. After the indexer is created (or by editing it), set the following property so it runs in the private execution environment:

      "parameters": {
        "configuration": {
          "executionEnvironment": "private"
        }
      }
   

   Without this setting the indexer still runs on the multi-tenant plane and attempts a public connection, resulting in the 403 error.
4. Ensure the search service managed identity has the required RBAC roles (Storage Blob Data Reader, Cognitive Services User / OpenAI User, etc.).
5. Vectorize and enrich your images” – no models appear
   The wizard only shows models that support image (multimodal) vectorization. Text-only models will never appear in that section.

Supported models for images (official list):

Recommendations:

• Deploy either Azure Vision multimodal (via a multi-service account) or a serverless Cohere-embed-v3 model in a hub-based project.
• Create the corresponding shared private link from Azure AI Search if public access is disabled.

Reference:
https://learn.microsoft.com/en-us/azure/search/search-indexer-howto-access-private?tabs=portal-create
https://learn.microsoft.com/en-us/azure/search/search-indexer-securing-resources
https://learn.microsoft.com/en-us/azure/search/search-get-started-portal-import-vectors?tabs=storage-access%2Cblob-storage%2Caoai%2Cvectorize-images

Kindly let us know if you have any furthr queries.

Please "upvote" if the information helped you. This will help us and others in the community as well.