I had a lot of fun this morning, brainstorming and assembling this Analytics Pipeline for Dashboards with Claude (and yes it takes much, much more than only 10 prompts).

The philosophy is borrowed from Observable Framework’s data loader, but implemented from scratch with voluntarily opinionated choices of libraries for Python, R and Javascript (and no Markdown): pre-compute data with Python + Polars (Polars is my new crush) or R + dplyr, publish it as static files, and explore them with JavaScript in the browser (Arquero for data wrangling, Observable Plot or Highcharts for charts)

The repository is available on GitHub and here’s the quick start guide to run it locally (you need to have Python and R installed on your machine):

make                        # list all targets
uv venv venv                # or: make venv (python -m venv venv without uv)
source venv/bin/activate    # Windows: venv\Scripts\activate
make install                # Polars into venv/, dplyr into .uvr/library/
make dev                    # starts http://localhost:3000 and opens your browser

The repository’s README file is also informative and contains a few more details about the pipeline.