Blog header image

Consistent Branding of reports and applications with LLMs

The analytics team at Presage uses tools such as Quarto and Shiny for developing reports, dashboards, and applications. These outputs require significant customization based on the company branding guidelines. The brand.yml project by Posit makes the branding process easier by requiring a single _brand.yml file that is placed in your project folder. This file should contain information from the branding guidelines, including fonts, colours, and logos. The accompanying brand.yml R and Python packages also bundle functions for theming visualizations and tables for a consistent look of reports and applications, making the branding process significantly faster.

For further increase in productivity, we can leverage three features of Large Language Model (LLM) APIs for generating _brand.yml files and colour palettes in our projects. These features are (1) Tool calling, (2) Retrieval-Augmented Generation (RAG), and (3) Structured output.

The following functions are bundled in the R package brandthis:

FunctionDescriptionLLM Feature(s)
create_brandCreates _brand.ymlImage-to-colour extraction, contrast checking
suggest_color_scalesSuggests suitable color palettesRAG
create_color_paletteCreates color palettesStructured output

Check out this presentation about brandthis below or view full screen here.

Last modified: December 11, 2025