AI Driven Coding Workflow - AI Discussions - DeepLearning.AI
The agentic multi-agent tools often have generic workflows that do not integrate well with other processes. In response to this, I have developed a unique software development workflow that operates at a file level, making direct calls to APIs to translate code and create tests to verify requirements.
I am particularly enthusiastic about the potential of GPT-powered technologies to take an outline and automatically generate code. Through my workflow, which includes a Python script and Makefile, I have streamlined the development process by enabling changes to the code to be easily back-propagated into the overall use case.
Currently, the Python script and Makefile in my workflow assemble inputs, prompting the OpenAI API to automatically generate code and run unit tests. Each iteration of this process only takes a few seconds, thanks to the efficient autogen code feature. To ensure smooth translation, it is necessary to break the code into smaller sections to stay within the token limit and reduce the time needed for translation.
For more discussions on AI and coding workflows, visit Discourse. Enable JavaScript for the best viewing experience.
