LAPLACE: una plataforma de IA para usarse con múltiples repositorios de código de proyectos de software

The emergence of large language models (LLMs) can be applied to the optimization and improvement of source code within software project development environments, providing automatic feedback and contextual suggestions, and leveraging the project repository itself as a source of knowledge to support the programming process. LAPLACE proposes the design and implementation of a unified platform for integrating large language models focused on software development. The proposed solution avoids the fragmentation resulting from the use of multiple independent environments. A key feature of LAPLACE is its high versatility regarding the large language models involved: it allows interaction with both global APIs (OpenAI, Anthropic, Google) and the deployment and use of local models (such as Ollama) by default. This flexibility ensures proper management of sensitive information, reducing or eliminating the exposure of source code to third-party services. To optimize workflows, LAPLACE implements an architecture based on Retrieval-Augmented Generation (RAG) and AI Agents. This approach uses code repositories as a contextual knowledge base, enabling the model to provide assistance, analysis, and precise feedback tailored to the project’s codebase. This promotes change traceability, data security, and efficient collaboration among development teams.