Dual-LLM Prompt Refinement for Lower-Effort Technical Problem Solving
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual prompt engineering for Large Language Models (LLMs) is labor-intensive and time-consuming, requiring constant updates to adapt to language and user behavior changes, and often results in unsatisfactory outputs due to unclear or contradictory prompts.
Innovation Solution
An automated method using two LLMs to generate improved task-specific prompts based on user interactions and feedback, where initial information is stored and used to formulate a new prompt for better LLM responses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If manual prompt engineering is used to optimize LLM responses, then response quality can be improved, but labor intensity and time consumption increase significantly
Solution Approach 1:
The system enables automatic prompt optimization where the LLM itself analyzes user interactions and generates improved prompts without requiring manual intervention. The automated prompt engineering system processes user queries, evaluates responses, and iteratively refines prompts using the LLM's own capabilities, making the system self-improving and eliminating the need for human prompt engineers.
Solution Approach 2:
The system implements feedback loops where user interactions with the LLM are continuously monitored and analyzed. The automated system uses this feedback to identify patterns, evaluate response quality, and generate optimized prompts for future queries. This closed-loop feedback mechanism enables continuous improvement of prompt effectiveness without manual intervention.
2Reliability
If manual prompt updates are performed to adapt to language changes, then LLM performance can be maintained, but labor intensity increases
Solution Approach 1:
The system automatically monitors language evolution and user behavior patterns, using the LLM's own analytical capabilities to identify linguistic changes and optimize prompts accordingly. The automated prompt engineering system performs self-updates without requiring human intervention, maintaining LLM performance while eliminating manual labor.
Solution Approach 2:
The system performs preliminary analysis of user interactions and language patterns to proactively generate optimized prompts before performance degradation occurs. By continuously monitoring and preparing updated prompts in advance, the system maintains optimal LLM performance without requiring reactive manual updates.
3Measurement precision
If clear and precise prompts are formulated, then LLM response accuracy improves, but prompt development complexity increases
Solution Approach 1:
The automated prompt engineering system uses the LLM's own capabilities to analyze user queries, evaluate potential responses, and generate optimized prompts automatically. This self-service approach eliminates the need for complex manual prompt development while maintaining high response accuracy through the LLM's inherent language understanding and optimization capabilities.
Solution Approach 2:
The system introduces an automated intermediary layer between the user's simple query and the LLM's response generation. This intermediary automatically handles the complexity of prompt formulation, translation, and optimization, allowing users to provide simple inputs while the system manages the complex prompt engineering required for accurate responses.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a computer-implemented method for supporting the solution of a technical problem by means of a first Large Language Model (B), in which a prompt (E) pertaining to the technical problem is provided to the first Large Language Model (B) from a database (A) for storing prompts (E). Initial information (IA) concerning a user's interaction with the first Large Language Model (B) in the context of solving the technical problem is stored and provided to a second Large Language Model (D). The second Large Language Model (D) then generates a newly formulated prompt (E-UP) for the technical problem.