ERP Query Response Using Prompt-Guided Generative AI
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing generative AI models, pretrained on public data, lack understanding of specific databases in ERP systems, leading to misaligned outputs and potential data corruption risks.
Innovation Solution
Integrate prompt engineering to dynamically retrieve data and metadata from ERP system databases, embedding this information into prompt templates to guide generative AI responses, ensuring contextual relevance and adherence to data policies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If existing generative AI models pretrained on public data are integrated into ERP systems, then automation and content creation capabilities are enhanced, but the models lack understanding of specific ERP database information leading to misaligned outputs
Solution Approach 1:
The system performs preliminary actions by extracting and storing metadata about ERP database tables, columns, and relationships before generating AI responses. This pre-processing creates a knowledge base that guides the generative AI to produce outputs aligned with specific ERP data structures and business context.
Solution Approach 2:
An intermediary layer is introduced between the generative AI model and the ERP database. This layer includes components that extract metadata, create prompt templates, and validate outputs, serving as a mediator that ensures the AI understands and correctly interprets ERP-specific data structures while maintaining automation capabilities.
2Device complexity
If generative AI models are integrated into ERP systems without contextual information, then implementation complexity is reduced, but data integrity risks increase due to potential data corruption
Solution Approach 1:
The system implements feedback mechanisms where AI-generated outputs are validated against ERP data policies and constraints before being executed. The metadata extraction and prompt template validation provide feedback loops that prevent data corruption by ensuring AI actions comply with ERP system rules and data integrity requirements.
Solution Approach 2:
The system prepares protective measures in advance by extracting and storing metadata about data policies, constraints, and validation rules before AI operations are performed. This pre-prepared knowledge acts as a cushion that prevents harmful AI actions from corrupting ERP data, reducing implementation complexity while maintaining safety.
3Measurement precision
If metadata extraction and prompt template generation are implemented, then contextual relevance of AI responses is improved, but system processing time increases
Solution Approach 1:
Metadata extraction and prompt template generation are performed as preliminary actions that cache contextual information for future use. By pre-processing and storing ERP database metadata and relationship information, the system reduces the processing time required for subsequent AI queries while maintaining high contextual relevance.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A computer-implemented method for intelligent query response in an enterprise resource planning (ERP) system is disclosed. The method can receive, from a user interface of the ERP system, a natural language user query for processing data of a target object and invoke a function call through an application programming interface of the ERP system based on the user query. Invoking the function call includes extract denormalized data from one or more database tables, generating a prompt using a prompt template and the denormalized data, prompting a large language model using the prompt, and receiving a response generated by the large language model. The method can generate an output on the user interface based on the response generated by the large language model. Related systems and software for implementing the method are also disclosed.