Custom Data Preparation Pipelines
Feed third-party data into Data Pools (RAG) through preparation pipelines
For complex integrations, we support a pipeline pattern with S3 as the handover point.
Hosting model
- Either the customer uses their own S3 bucket
- Or meinGPT can host an S3 bucket on request
Typical workflow
- Data is extracted from third-party systems (e.g., DMS, ERP, CRM, or domain tools)
- A script or flow prepares the content (e.g., into document-ready formats)
- Prepared documents are written into the S3 bucket
- meinGPT ingests the bucket and exposes the content through RAG
For extraction and preparation, no-code flows such as n8n can be used.
Security
Transfer and storage can be implemented with encryption (e.g., SSE-C or comparable customer-managed encryption), depending on security requirements.