poppychat provides an R code interface to chat with large language models through Poppy, a service provided by the California Department of Technology in a pilot to a group of State of California employees. Poppy models are sandboxed for State use. poppychat is intended to facilitate the secure integration of Poppy’s capabilities into R code, dashboards, and other applications.
State of California employee use of Poppy should adhere to the latest GenAI State Policies.
Installation
You can install the development version of poppychat from GitHub with:
devtools::install_github("lucy-dwr/poppychat")Usage
poppy_chat(), the main function in this package, creates a chat client configured to connect to Poppy using a user-supplied API key. poppy_chat()is a thin wrapper for ellmer::chat_openai_compatible(). To get started, set the POPPY_API_KEY environment variable. See vignette("poppychat") for details.
gemini_chat <- poppychat::poppy_chat(model = "Google Gemini 2.5 Flash")
gemini_chat$chat("What's the best thing about working for the State of California? Be brief!")
#> Many employees value the opportunity to serve the public and make a
#> a meaningful difference for Californians, alongside comprehensive benefits
#> and job stability.ellmer chat clients support a variety of features, including:
- Conversation history
- Token tracking
- System prompts
- Tool calling
- Structured data extraction
- Retrieval-augmented generation
For more information on using a chat client, review package documentation from ellmer.