Prodigy - S-Drive Integration and Automation

Common Integration Use Cases Between Prodigy and S-Drive

Prodigy and S-Drive complement each other well in organizations that need to manage training data, supporting documents, and review workflows across AI, operations, and compliance teams. Prodigy handles structured data annotation for machine learning, while S-Drive provides secure document collection and storage inside Salesforce. Together, they can streamline how source content is gathered, reviewed, labeled, and governed.

1. Collect source documents in Salesforce and send them to Prodigy for annotation

Direction: S-Drive to Prodigy

Business teams can use Salesforce cases, opportunities, or custom records to collect documents such as contracts, invoices, claims, forms, or product images through S-Drive. Once documents are approved or ready for AI training, they can be exported to Prodigy for labeling by data science or operations teams.

• Salesforce users upload files directly to a record using S-Drive
• Relevant files are routed to Prodigy for image, text, or document annotation
• Annotated outputs are used to train custom models for classification, extraction, or detection

Business value: Reduces manual file handling, speeds up dataset creation, and keeps source documents tied to business records.

2. Return labeled outputs from Prodigy to Salesforce for operational review

Direction: Prodigy to S-Drive

After annotation, Prodigy can generate labeled files, JSON exports, or review artifacts that are stored back in S-Drive and linked to the originating Salesforce record. This is useful when business users need to validate training results or retain labeled evidence for audit purposes.

• Annotated datasets are exported from Prodigy after review
• Outputs are stored in S-Drive against the related Salesforce case or account
• Stakeholders can access both original and labeled versions from Salesforce

Business value: Improves traceability, supports audit readiness, and gives business teams visibility into AI training inputs and outputs.

3. Use Salesforce case workflows to trigger labeling tasks in Prodigy

Direction: S-Drive to Prodigy

When a Salesforce case reaches a specific status, such as ?ready for model training? or ?needs review,? the associated documents in S-Drive can be sent to Prodigy for annotation. This is especially useful in customer support, claims processing, and compliance operations where new examples must be labeled continuously.

• A case status change triggers document transfer to Prodigy
• Labeling tasks are assigned to internal reviewers or subject matter experts
• Completed annotations are used to improve AI models for automation or decision support

Business value: Creates a repeatable workflow from business event to AI training, reducing delays between operations and model improvement.

4. Store annotation guidelines and reference materials in S-Drive for Prodigy reviewers

Direction: S-Drive to Prodigy

Organizations often need controlled access to labeling instructions, policy documents, and reference examples. S-Drive can manage these files in Salesforce, while Prodigy users access them as supporting material during annotation. This helps maintain consistency across distributed labeling teams.

• Labeling guidelines are stored in S-Drive and linked to a Salesforce record or project
• Prodigy annotators access the latest approved instructions before labeling
• Updates to guidelines in Salesforce are reflected in the annotation process

Business value: Improves label quality, reduces rework, and ensures annotators follow the same business rules.

5. Support human-in-the-loop review for AI-assisted document processing

Direction: Bi-directional

In document-heavy workflows, an AI model may pre-classify or extract fields from files stored in S-Drive, and Prodigy can be used to review uncertain predictions or edge cases. Corrected labels can then be stored back in Salesforce for downstream business processing.

• Documents are collected in S-Drive and processed by an AI model
• Low-confidence results are sent to Prodigy for human review and correction
• Validated outputs are written back to Salesforce records and document repositories

Business value: Increases automation while preserving accuracy for exceptions and complex cases.

6. Build and maintain training datasets from customer-submitted documents

Direction: S-Drive to Prodigy

Customer-submitted files stored in Salesforce, such as onboarding forms, identity documents, or service requests, can be periodically extracted from S-Drive and labeled in Prodigy to create training datasets for OCR, document classification, or entity extraction models.

• Documents are collected through Salesforce portals or internal workflows
• Selected samples are exported to Prodigy for structured labeling
• Trained models are used to automate future document handling

Business value: Helps teams turn operational documents into reusable AI assets and improves processing efficiency over time.

7. Maintain compliance evidence for model training and annotation governance

Direction: Prodigy to S-Drive

For regulated industries, it is important to retain evidence of what data was used, who labeled it, and which version of the dataset supported a model release. Prodigy outputs, review logs, and labeling artifacts can be stored in S-Drive alongside Salesforce records for governance and audit support.

• Annotation exports and review history are archived in S-Drive
• Files are linked to project, case, or compliance records in Salesforce
• Audit teams can retrieve evidence without accessing the annotation environment directly

Business value: Strengthens governance, supports regulatory audits, and improves accountability across AI programs.

8. Route approved document sets from Salesforce to Prodigy for active learning cycles

Direction: S-Drive to Prodigy

Prodigy?s active learning approach works best when it receives the most informative samples. Salesforce can act as the system of record for approved document sets, while S-Drive supplies new or changed files to Prodigy based on business rules such as document type, region, or customer segment.

• New documents are stored in S-Drive and filtered by Salesforce criteria
• Prodigy selects high-value samples for labeling and model improvement
• Results are fed back into the AI pipeline for the next training cycle

Business value: Focuses labeling effort on the most valuable data, reducing annotation cost and improving model performance faster.