Agentic and generative AI features in Pega Smart Investigate Agentic Automation
Payment investigations often rely on email communication. Traditional approaches struggle to process unstructured content and automate analysis efficiently. Pega Smart Investigate Agentic Automation™ (SIAA) addresses these challenges by introducing advanced natural language processing (NLP) and agentic AI features that extract key information, determine intent, and route investigations with institution-specific accuracy.
These AI-powered features enable financial institutions to automate email processing while maintaining flexibility to tune models for unique terminology, customer phrasing, and investigation scenarios.
NLP model
Email remains a primary channel for payment investigations, which requires efficient handling of unstructured content. Previous solutions relied on basic email integration with limited automated analysis. SIAA introduces an advanced NLP model that analyzes incoming emails to:
- Extract key information.
- Determine intent.
- Route investigations appropriately.
You can tune the NLP model to recognize institution-specific terminology, common customer phrasing, and unique investigation scenarios to improve automated processing accuracy.
Review the following process for configuring the NLP model in your application:
- Collect a representative sample of investigation emails from current operations.
- Include emails from different investigation types.
- Ensure the representation of various customer segments and regions.
- Include both simple and complex investigation scenarios.
- Configure the NLP model.
- Train the system to recognize institution-specific terminology.
- Define intent recognition patterns for common investigation types.
- Set confidence thresholds for automated processing versus manual review.
- Implement a feedback loop to improve accuracy.
- Track false positives and false negatives.
- Update the model regularly with new examples.
Agentic and generative AI features
SIAA represents the next evolution in investigation operations, moving beyond automation to intelligent, learning-optimized outcomes. Several integrated agentic capabilities fundamentally re-imagine how payment investigations are managed, combining autonomous decision support, process orchestration, and revenue optimization.
SIAA introduces the following agentic features:
- Smart Investigation Agent
- Accounting Agent
- Correspondence Agen
- Compensation Agent
- ProcessAI 'Next Best' Resolution activity
- Agentic Email
SIAA introduces the following generative AI features:
- Chat to My Data
- Pega Process Mining
- Pega GenAI Knowledge Buddy™
- Pega GenAI Coach™
Review the following process for implementing agentic features:
- Add agentic features based on organizational AI strategy and licensing availability.
- Tune models and NLP configurations as needed.
- Ensure that data is available for reporting, analytics, and process mining.
- Populate the vector database with standard operating procedures and documents to enable Knowledge Buddy to operate effectively.
This Topic is available in the following Module:
Want to help us improve this content?