Programming with an AI agent
Artificial intelligence (AI) has fundamentally changed how we program. AI agents can generate code, optimize it and even assist with debugging. Still, there are limitations programmers must keep in mind when working with AI.
Issues with ordering and duplication
AI agents struggle with correct code ordering. For example, they may place initializations at the end of a file, causing runtime errors. Additionally, AI can confidently define multiple versions of the same class or function within a project, leading to conflicts and confusion.
A code platform with memory and project structure helps
One remedy is using AI-code platforms that manage memory and project structure. This helps maintain consistency in complex projects. Unfortunately, these features are not always applied consistently, so the AI can lose track of a project’s coherence and introduce unwanted duplications or incorrect dependencies during development.
Most AI coding platforms work with so-called tools that can invoke the large language model. Those tools are based on an open standard protocol (MCP). It is therefore possible to connect an AI coding agent to an IDE such as Visual Studio Code. Optionally, you can set up a local LLM with llama or ollama and choose an MCP server to integrate with. NetCare has a MCP server designed to help with debugging and managing the underlying (Linux) system. Useful when you want to deploy code live immediately.
Models can be found on Hugging Face.
IDE extensions are indispensable
To better manage AI-generated code, developers can use IDE extensions that enforce code correctness. Tools like linters, type checkers and advanced code analysis utilities help detect and correct errors early. They are an essential complement to AI-generated code to ensure quality and stability.
The cause of recurring errors: context and role in APIs
One of the main reasons AI agents keep repeating mistakes lies in how AI APIs are interpreted. AI models need context and a clear role definition to generate effective code. That means prompts must be complete: they should not only include functional requirements but also explicitly state expected outcomes and constraints. To facilitate this, you can store prompts in a standard format (MDC) and always send them with requests to the AI. That is particularly useful for generic programming rules you follow and for the functional and technical requirements and structure of your project.
Tools such as FAISS and LangChain help
Products such as FAISS and LangChain offer solutions to help AI work better with context. FAISS, for example, assists with efficient search and retrieval of relevant code snippets, while LangChain helps structure AI-generated code and preserve context within a larger project. You can also set this up locally with RAC databases.
Conclusion: useful, but not yet autonomous
AI is a powerful tool for developers and can help accelerate development processes. However, it is not yet capable of independently designing and building a complex codebase without human oversight. Developers should treat AI as an assistant that can automate tasks and generate ideas, but still requires guidance and correction to produce a good result.
Contact contact to help set up the development environment so teams can get the most out of it and focus more on requirements engineering and design than on debugging and writing code.
