The integration process with Wasabi AiR consists of the following high-level steps:
Enable AiR access for the desired Wasabi Account and users.
Initiate analyses by creating jobs on storage objects using our Public AiR API. A job is created with authentication, input storage object, output storage location for generated metadata and desired AI services to be executed. Refer to the API documentation link for details.
Monitor jobs for status updates or subscribe to filesystem events for status updates.
When the job completes, the resulting generated metadata is written to the desired account bucket location.
The resulting metadata is written out as a JSON storage object in the customer’s account, where the customer owns the data.
The persisted metadata is then parsed for information, indexed by an external system such as Elasticsearch or other indexes, or mapped into a MAM/DAM for discovery of managed assets by the archive owner.
We have purposely designed the process to initiate a job that generates or persists the metadata as easily as possible, where much of the integration involves mapping that metadata to external systems.
Monitoring jobs can be managed by polling the Job API for individual job status or by subscribing to filesystem events to generate metadata JSON storage objects.
Configuring Object Event Notifications
The API documentation clearly explains how to use a service configuration. A service configuration allows a client to configure which ML services are executed against the storage object.
Failed jobs display an error message that explains the reason for the failure. Most of those failures are caused by incorrectly encoded or formatted files. Ensure the files adhere to the supported video, audio, and image codecs and only .pdf documents. If no audio was analyzed, it is because the incorrect audio channels were configured in the service config.