In Pega 25.1.X or 26.1, I would like to get the validate whether Pega GenAI can process and extract information and present it to case from:
Standard document attachments such as DOCX
PDF attachments.
Multiple attachments ( combination of PDF, Images, DOCX)
Large attachments.
ZIP files containing multiple documents.
- What are the preferred implementation approach + design patterns / best approaches to consider & implement these types of attachments based on available Gen AI rule types?
- What are the limitations / constraints that we need to consider?
Looking forward to hear your experiences / inputs / alternatives to tackle this enterprise document heavy challenge.
One remark about file sizes: please keep in mind that attachments will get converted to Base64, which increases their size. An image that is 3.75 MB on disk is around 5 MB as Base64.
As far as I know, Bedrock endpoints only handle attachments up to 5 MB (Base64).
About your question around multiple attachments: you can manually ingest multiple attachments to GenAI agents; however, GenAI Connect rules only support single attachment analysis. Check out DocAI with Custom Attachment Lists if you want further details on this and a very interesting workaround by Fergal.
