Skip to content

AI Disclosure

AI was used during the development of Quicksend. It is my believe that being transparent about the usage of such tool is important.

For the purposes of this project, Artificial Intelligence (AI) refers exclusively to Large Language Models (LLMs). Quicksend strictly utilized LLMs (such as ChatGPT, Gemini, and Mistral). No generative diffusion models, image synthesis tools, or autonomous agents were employed in the development of this work.

Goals

The use of AI in this project serves to assist and not replace human knowledge and judgment. Human developers remain 100% responsible for the security, performance, and maintenance of the final product. Every AI-assisted suggestion must be subjected to rigorous manual verification before implementation.

Permitted usages

Usage of LLMs was restricted to the following areas:

  • Research: Getting an overview of technologies, researching capabilities, bad practices
  • Code generation: LLMs were sometimes used to generate code like pipeline scripts. LLM generated code is always human reviewed before being merged.
  • Code review: LLMs were sometimes used to identify obvious logic flaws or deviations from best
  • Documentation: "Blockouts" of the documentation are generated with LLMs. These first drafts however - as with all parts of this project - are human- reviewed, rephrased and checked.

Forbidden usages

To maintain the integrity and knowledge of the codebase, the following restrictions are enforced:

  • System architecture: All higher-level structural decisions, database schemas, and system designs were conceived and finalized solely by humans
  • No direct contributions: Code conceived directly by AI/LLMs is not permissible. Every line of code within the repository has been manually implemented, reviewed, and merged by a human developer.
  • Localization: LLMs are not used to generate primary translations or localization strings. While AI may have been used to verify the overal validity and formatting of localization files, all linguistic content remains human-authored to ensure cultural accuracy.