Prodigy - OpenText Workflow Service Integration and Automation

Common Integration Use Cases Between Prodigy and OpenText Workflow Service

1. Human-in-the-loop data labeling approval workflow

Flow: Prodigy ? OpenText Workflow Service

When data scientists complete an annotation batch in Prodigy, the labeled dataset can be sent to OpenText Workflow Service for formal review and approval by business SMEs, compliance teams, or model owners. OpenText can route the task to the right approvers, capture sign-off, and maintain an auditable record before the dataset is released for model training.

Business value: Improves governance over training data, reduces labeling errors, and creates accountability for regulated or high-impact AI use cases.

2. Exception handling for low-confidence or disputed labels

Flow: Prodigy ? OpenText Workflow Service

Items flagged in Prodigy as ambiguous, low-confidence, or disputed by annotators can be automatically escalated into an OpenText case or workflow. The workflow can assign the item to senior reviewers, subject matter experts, or quality assurance teams for resolution, then return the final decision back to Prodigy.

Business value: Reduces rework, accelerates resolution of difficult labels, and ensures consistent handling of edge cases.

3. Controlled release of approved datasets into MLOps pipelines

Flow: Prodigy ? OpenText Workflow Service ? downstream systems

After annotation is completed in Prodigy, OpenText Workflow Service can orchestrate a release process that validates dataset completeness, checks required approvals, and confirms policy compliance before the dataset is handed off to training or MLOps platforms. This is especially useful when datasets must meet internal controls before model retraining.

Business value: Prevents unapproved data from entering production AI pipelines and strengthens operational control over model development.

4. Annotation request intake and prioritization

Flow: OpenText Workflow Service ? Prodigy

Business teams can submit new labeling requests through OpenText Workflow Service, including source content, labeling instructions, priority, due date, and required approvers. Once approved, the workflow can create a corresponding labeling project in Prodigy and assign it to the appropriate annotation team.

Business value: Standardizes demand intake, improves prioritization across departments, and gives AI teams a structured request process.

5. Compliance-driven labeling for regulated content

Flow: Bi-directional

For regulated industries such as healthcare, financial services, or insurance, OpenText Workflow Service can manage policy checks, access approvals, and case documentation before sensitive content is exposed in Prodigy. After labeling, Prodigy can send completion status and annotation metadata back to OpenText for audit trails and case closure.

Business value: Supports traceability, access control, and audit readiness for sensitive datasets.

6. Quality assurance and rework loop for annotation operations

Flow: Prodigy ? OpenText Workflow Service ? Prodigy

Quality reviewers can use OpenText Workflow Service to manage structured QA checks on completed annotation batches from Prodigy. If issues are found, the workflow can route the batch back to Prodigy with correction instructions, reviewer comments, and required rework actions.

Business value: Creates a repeatable QA process, improves label consistency, and reduces downstream model training defects.

7. Dataset lineage and project governance reporting

Flow: Prodigy ? OpenText Workflow Service

Prodigy can provide annotation status, reviewer decisions, and dataset version details to OpenText Workflow Service, which can store the workflow history alongside related business cases and content records. This gives stakeholders a single place to track who approved what, when labeling occurred, and which dataset version was used for a model release.

Business value: Strengthens dataset lineage, supports auditability, and improves transparency across AI delivery teams.

8. Cross-functional model improvement request management

Flow: OpenText Workflow Service ? Prodigy

Operations, customer service, or compliance teams can submit model improvement requests in OpenText Workflow Service when they identify new data patterns, errors, or policy changes. Approved requests can trigger new annotation tasks in Prodigy so the AI team can quickly gather fresh training data and retrain models.

Business value: Shortens the feedback loop between business users and AI teams, helping models adapt faster to changing business needs.