How to adopt agentic coding in a large Python codebase?
agentic developmentllmProgramming
I'd like to shift our development workflow to an agentic approach, enabling us to use LLM models more efficiently in actual software development. We are a team of very senior experts in a Python-heavy niche industry. Currently, our adoption level is low. We are stuck asking for snippets and help, and have not managed to get agents to build larger modules beyond disposable scripts.
Synthesized Answer
Based on 1 community responses
To transform your development workflow to an agentic approach, start by setting up a structured codebase that agents can navigate. This involves organizing your files and installing necessary hooks for agent integration. Utilize tools like CLI/SaaS that can enhance agent intelligence, such as LangChain or LlamaIndex, to improve their ability to understand and work with your codebase. When selecting a model, consider factors like context size, training data, and fine-tuning capabilities to ensure it aligns with your project's needs. Write clear, concise instructions in a format the agent can understand, and establish a 'loop' where the agent can work independently, receiving feedback and guidance as needed. You can also explore using multiple agents to tackle different aspects of your project, improving overall efficiency and productivity.
Key Takeaways
Set up a structured codebase for agent navigation
Utilize tools like LangChain or LlamaIndex to enhance agent intelligence
Select a suitable model based on context size, training data, and fine-tuning capabilities
Write clear instructions in a format the agent can understand
Establish a 'loop' for independent agent work with feedback and guidance