LLM Process Pilot for Actionable BPM Error Resolution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Business process management (BPM) application users face challenges in accessing necessary information within the application's user interfaces, requiring manual searches that are time-consuming and burdensome, especially for managers and executives with limited time.
Innovation Solution
Implementing a process pilot tool that leverages a large language model (LLM) to automatically generate actionable information by collecting and vectorizing static and instance-specific data, detecting errors, and providing natural language outputs through a chatbot-like interface within BPM application dashboards.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If manual searching is used to find information, then users can access information from multiple sources, but the process is time-consuming and burdensome
Solution Approach 1:
The system enables self-service by automatically collecting, vectorizing, and organizing information from multiple sources without requiring users to manually search. The process pilot tool autonomously retrieves static and instance-specific information, transforms it into actionable insights, and presents them directly to users within the BPM application interface, eliminating the need for manual information gathering while maintaining comprehensive information accessibility.
Solution Approach 2:
The system performs preliminary action by proactively collecting and processing information before users need it. The process pilot tool continuously gathers static information (help documentation, troubleshooting guides) andinstance-specific information (logs, process visibility events, uploaded documents), vectorizes them, and stores them in knowledge bases ready for immediate retrieval when users encounter issues or need information.
2Loss of information
If managers and executives perform manual searching, then they can access detailed information, but their limited time is consumed
Solution Approach 1:
The system enables self-service by automatically collecting, vectorizing, and organizing information from multiple sources without requiring users to manually search. The process pilot tool autonomously retrieves static andinstance-specific information, transforms it into actionable insights, and presents them directly to users within the BPM application interface, eliminating the need for manual information gathering while maintaining comprehensive information accessibility.
Solution Approach 2:
The system implements feedback by detecting errors and process visibility threshold violations, then automatically retrieving relevant information and generating actionable insights that are presented to users. This closed-loop approach ensures that when managers or executives encounter issues, the system provides targeted information and suggested actions based on real-time error detection and knowledge base retrieval, improving their efficiency without compromising information completeness.
3Productivity
If a process pilot tool with LLM is implemented, then information access is streamlined and error resolution is accelerated, but device complexity increases
Solution Approach 1:
The system uses an intermediary approach by introducing a process pilot tool that acts as a mediator between users and the complex information systems. This tool includes an LLM-based interface that simplifies interactions with static andinstance-specific information, automatically handles error detection, retrieves relevant knowledge, and presents actionable insights. The intermediary layer abstracts the complexity of information collection, vectorization, and processing from the user interface, maintaining high productivity while managing system complexity through a dedicated specialized component.
Data Source
AI summary
Techniques for automatically generating actionable information pertaining to business process instances using large language models (LLMs) are provided. In certain embodiments the information generated via these techniques can support participants of the business process instances, such as task owners and process managers, in taking actions and making decisions on the instances in accordance with their respective responsibilities.