AI Software Workflow Tool for Testable Agile Development
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing software development tools prioritize speed over quality and reliability, leading to increased technical debt, project cost overruns, and unmet project goals, while lacking integration with development methodologies and versatility in generating comprehensive development assets.
Innovation Solution
GILES, a software development tool leveraging Generative AI, integrates Agile and Behavior-Driven Design (BDD) best practices to generate testable requirements, specifications, test cases, and documentation, emphasizing rigorous testing and customized AI prompting to ensure high-quality, maintainable code.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If traditional software development tools are used to prioritize speed, then development speed is improved, but code quality and reliability deteriorate
Solution Approach 1:
The system implements feedback loops where AI models generate code, then automatically test and evaluate the generated code quality. Test results and quality metrics are fed back to refine subsequent code generation, ensuring continuous improvement of reliability while maintaining speed through automated evaluation rather than manual review.
Solution Approach 2:
The system performs preliminary actions by generating comprehensive test cases, requirements specifications, and acceptance criteria before actual code development. This upfront planning and testing framework ensures that code quality standards are established in advance, preventing reliability issues rather than detecting them after slow manual review.
2Reliability
If comprehensive software development assets are generated with rigorous testing, then code quality is improved, but development time increases
Solution Approach 1:
The system replaces manual mechanical processes of code review, test case creation, and quality assurance with AI-driven automated systems. Machine learning models generate code, automatically create and execute test cases, and perform quality evaluation, dramatically reducing the time required to achieve comprehensive code quality standards compared to traditional manual processes.
Solution Approach 2:
The system maintains continuous automated testing and quality validation throughout the development process rather than performing these activities in separate batch phases. This continuous integration of AI-based testing and quality assurance ensures code quality is maintained without interrupting the development flow, eliminating the time loss associated with stopping for comprehensive quality checks.
3Productivity
If AI models are used to generate software assets, then productivity is improved, but resource consumption increases
Solution Approach 1:
The system applies partial action by using AI models selectively for specific high-value tasks such as generating test cases, creating requirements specifications, and performing code review, rather than attempting to use AI for every aspect of development. This targeted approach achieves significant productivity improvements while limiting computing resource consumption to only the most beneficial applications.
Solution Approach 2:
The system dynamically adjusts AI model parameters such as generation length, complexity levels, and testing depth based on project requirements and resource availability. By changing these parameters, the system can optimize the balance between productivity gains from AI assistance and the computing resources consumed, scaling AI usage to match available resources while maintaining effective productivity improvement.
Data Source
AI summary
Systems and methods for software development tool and use thereof. The method can include accessing input data including a config file. The method can include processing the config file to generate output including (i) phases of a workflow and (ii) steps for the respective phases of the workflow. The method can include initiating a chat session for a step. The chat session including a conversation between a user and a first machine-learned model to work on the step. The method can include determining that the step has been completed and extracting output including conversation context data. The method can include determining that all phases of the workflow are complete. The method can include generating, based on the output and the conversation context data, output including an executable plan indicative of the one or more steps and the one or more phases.