Before
A reviewer had to retrieve the supplier attachment separately and prepare its contents for analysis. If the stored document was not readily usable, the sender could be asked to submit it again.
How it works
- Receive the filename and stored artifact references.
- Read the document through its binary and encoded paths.
- Submit the evidence for structured AI analysis and return the results.
Workflow details
- Receive the evidence reference: The workflow starts with a document filename and references to its binary and base64-stored versions.
- Read the stored document: It retrieves the evidence through both available representations so the review can use the referenced artifact rather than requiring a new upload.
- Analyze supplier posture: An AI analyst reviews the document as part of a supplier security assessment and extracts factual information about the document type, date, scope, auditor, validity, summary, key points, citations, and missing or unclear information.
- Standardize findings: The analysis is returned as a consistent JSON object with defined fields and controlled document-type and evidence classifications.
- Deliver reusable results: The workflow exposes the analysis through separate encoded and raw result outputs for use in the risk team’s supplier review.
What changed
The stored artifact reference now drives document retrieval and analysis through both supported representations, producing a structured supplier-evidence assessment without a separate resubmission handoff.
Work removed
Retrieving the stored document separately, repackaging its contents for review, and asking the sender to resubmit evidence that is already available.
Expected outcome
At least one repeated retrieval-and-repackaging cycle is removed from each supplier document review, leaving reusable structured evidence for the risk team.
