Get Started with aisdk

aisdk is a unified framework for building AI-powered applications in R. It abstracts the differences between providers (OpenAI, Anthropic) and provides a powerful set of tools for building agents, flows, and structured outputs.

Installation

You can install the development version from GitHub:

# install.packages("devtools")
devtools::install_github("YuLab-SMU/aisdk")

Basic Usage

1. Configuration

First, load the library and set your API keys as environment variables.

library(aisdk)

# RECOMMENDED: Set in your .Renviron file
Sys.setenv(OPENAI_API_KEY = "sk-...")
Sys.setenv(ANTHROPIC_API_KEY = "sk-ant-...")

2. Creating a Model

You can create a model provider using specific factory functions or the unified registry syntax.

# Option A: Factory Function
openai <- create_openai()
model <- openai$language_model("gpt-4o")

# Option B: Registry Syntax (if provider is registered)
registry <- get_default_registry()
registry$register("openai", create_openai())
# model <- "openai:gpt-4o"

3. Generating Text

Use generate_text() for simple request-response interactions.

response <- generate_text(
  model = model,
  prompt = "Explain the difference between lapply and sapply in R."
)

cat(response$text)

You can also use a system prompt to set the behavior:

response <- generate_text(
  model = model,
  system = "You are a concise data science instructor.",
  prompt = "What is a tibble?"
)

4. Streaming Responses

For real-time output, use stream_text().

stream_text(
  model = model,
  prompt = "Write a haiku about CRAN.",
  callback = function(chunk, done) {
    if (!done) cat(chunk)
  }
)

5. Switching Providers

Switching to Anthropic is as simple as changing the model creation step. The rest of your code remains the same.

# Create Anthropic provider
anthropic <- create_anthropic()
model <- anthropic$language_model("claude-3-5-sonnet-20241022")

# Generate text (SAME API!)
response <- generate_text(
  model = model,
  prompt = "Explain R6 classes briefly."
)
cat(response$text)