Using AI Agents

Prerequisites

The AI agent feature is only available in a Kubernetes setup.
To use PAS AI agents, you need an AI service provider account. This is not provided by Scheer PAS. Currently, OpenAI and Mistral are supported providers.

Integrating AI agents into your BPMN-based processes is a strong strategy, especially when processes require contextual understanding, dynamic decision-making, or heavy data processing. The use of AI agents can lead to significant improvements in the speed, quality, and adaptability of your business processes.

When to Use AI Agents?

AI agents can be helpful for a wide range of use cases.

• Improved Decision-Making: AI agents can evaluate large volumes of data or complex criteria to make decisions faster and more consistently than humans.
  Example Use Case: In a loan approval process, an AI agent can analyze credit scores, transaction history, and risk models to suggest or auto-approve/reject applications.

• Task Automation Beyond Rule-Based Logic: I can handle tasks that don’t follow deterministic rules, such as understanding language, recognizing images, or making contextual judgments.
  Example Use Case: Automatically routing support tickets based on content using natural language processing (NLP) instead of static keyword rules.

• Dynamic Process Adaptation: AI agents can dynamically alter the process path based on context, learning from past executions to optimize the process.
  Example Use Case: In supply chain workflows, an AI agent can reroute procurement or shipping paths based on delivery delays or inventory predictions.

• Human-AI Collaboration: Agents can act as co-pilots, assisting humans during steps in the process by providing recommendations, pre-filling forms, or summarizing information.
  Example Use Case: During an onboarding process, an AI agent helps HR by generating personalized welcome emails, summarizing candidate profiles, or suggesting next steps.

• Scalability & Efficiency: Once trained or configured, AI agents can run in parallel across thousands of process instances without fatigue or slowdown.
  Example Use Case: Handling inbound customer inquiries across multiple languages and regions via a multilingual AI assistant integrated into a customer support BPMN flow.

• Better Data Extraction and Structuring: AI agents (especially those powered by LLMs or computer vision) can extract structured data from unstructured inputs like PDFs, emails, or voice.
  Example Use Case: A document ingestion process uses an AI agent to read incoming invoices, extract key data, and validate them against PO records.

• Proactive Recommendations and Alerts: Agents can proactively monitor data or state across processes and suggest interventions before issues arise.
  Example Use Case: In a financial audit workflow, an AI agent flags anomalies or risky transactions for manual review before finalizing the process.

• Learning from Historical Process Data: AI agents can analyze process logs and suggest optimizations, bottleneck resolutions, or even redesigns of the BPMN model.
  Example Use Case: Process mining outputs feed into an AI agent that recommends altering task sequences to reduce cycle time.

How to use AI Agents in PAS?

We have integrated AI agents into PAS so that you can also profitably use the advantages of AI in your business processes. On the following pages, you will learn how it works:

Related Content