Built for RStudio

Your AI assistant, inside RStudio

AiR lives in your Viewer pane - an AI assistant that understands your entire R project alongside a complete data toolkit. Build models, explore data, generate reports, fix errors, and edit code - all without leaving RStudio.

Everything you need, right in RStudio

No context switching. No copy-pasting to ChatGPT. AiR understands your code, your data, and your errors - and works alongside you to fix, build, and explore.

Code Editing

Edit code with natural language

Describe what you want changed and AiR edits your active file directly. Refactor functions, add documentation, restructure code - all shown as a diff you can accept or reject.

  • Edits your active file in RStudio directly
  • Color-coded diffs with line-by-line changes
  • Accept or reject - with undo if you change your mind
  • Understands your full project context
Learn more about Code Editing
AiR - Viewer Pane
AiR
ChatDataTidyModel
clean_data.R
U
You
Add input validation to the clean_survey function
AiR
Reading clean_data.R...
Detected function:
clean_survey(df, threshold = 0.5)
df %>%
select(where(~mean(!is.na(.)) >= threshold))
# ... 12 more lines
AiR - Data Explorer
AiR
ChatDataTidyModel
diamonds53,940 x 10
Run EDACleanDescribe
2 cols with missing data price: 8% · 146 duplicate rows
caratcutcolorclaritydepthprice
0.23IdealESI261.5326
0.21PremiumESI159.8326
0.23GoodEVS156.9327
0.29PremiumIVS262.4334
0.31GoodJSI263.3335
Data Explorer

Explore your data and environment in real time

AiR syncs with your R session and gives you a live command center for your data. Preview rows, inspect columns, discover correlations, and take action - all while AiR understands every dataframe, file, and variable in your project.

  • Live sync with your R environment - every dataframe, column type, and distribution
  • Preview actual data rows in a scrollable table
  • Column inspector with types, ranges, missing data, and distributions
  • Correlations tab - top variable pairs ranked by strength
  • One-click Plot, Summarize, Tidy, and Run EDA actions
  • Knows your project file structure, active file, and selected code
Learn more about Data Explorer
Data Tidying

Clean, transform, and prepare your data

A dedicated workspace for data preparation. Audit your dataset for issues, apply transformations with one click, and chat with a tidyverse expert - all without touching the original data.

  • Automatic data quality audit: missing values, outliers, type issues, skewness
  • One-click transforms: impute, winsorize, log/sqrt, type conversions
  • Built-in tidyverse expert chat for joins, pivots, and custom operations
  • Never mutates your original dataset - all work on a safe copy
  • Export transforms as a reproducible .R script
  • Seamless handoff to Model Builder with your cleaned data
Learn more about Data Tidying
AiR - Data Tidying
AiR
ChatDataTidyModel
diamonds53,940 × 10
→ diamonds_tidy
4 issues found
diamonds · 53,940 rows × 10 columns
!
Outliers
price: 1,951 outliers (IQR method)
WinsorizeFlag
~
Skewness
price, carat: right-skewed (>2.0)
LogSqrt
A
Type Issues
cut, color, clarity: chr → factor
To Factor
=
Constants
No constant columns detected
Model Builder

Build and evaluate models interactively

Select a dataset, pick your variables, choose a model type, configure validation, and fit it live in R. Get structured results with metrics, coefficients, and confusion matrices - then an AI interpretation that cites actual values from your output.

  • 15 model types: LM, GLM (5 families), ridge, lasso, elastic net, logistic, naive bayes, KNN, decision tree, random forest, SVM, XGBoost, PCA, K-Means, hierarchical clustering
  • Supervised & unsupervised modes with interaction terms, stepwise selection, and class imbalance handling
  • Cross-validation: K-Fold, Repeated K-Fold, LOOCV - or classic train/test split
  • Diagnostic plots, model comparison table, and AI auto-fix for failed code
  • AI interpretation cites real values - no generic summaries
  • Export the exact R code as a clean, reproducible script
Learn more about Model Builder
AiR - Model Builder
AiR
ChatDataTidyModel
Model Builder
1
Dataset
2
Variables
3
Model
4
Results
5
Interpret
Select Dataset
mtcars
32 x 11
iris
150 x 5
diamonds
53,940 x 10
Next: Variables
AiR - Viewer Pane
AiR
ChatDataTidyModel
explore.R
U
You
/eda diamonds deep
AiR

Planning a deep EDA for diamonds (53,940 rows, 10 columns)...

Analysis Plan
1.Distribution of numeric variables (carat, price, depth, table)
2.Frequency analysis of categorical variables (cut, color, clarity)
3.Price correlations and interaction effects
4.Statistical tests (normality, ANOVA by cut)
5.Multivariate visualizations with faceting
EDA Reports

Full analysis reports in seconds

Say /eda and AiR generates a complete R Markdown report - distributions, correlations, statistical tests, and narrative insights. Choose quick, standard, or deep depth.

  • Generates complete .Rmd files with visualizations
  • Three depth levels: quick overview to full statistical analysis
  • AI plans the analysis strategy before generating
  • Validates R code and audits insights for accuracy
Learn more about EDA Reports
Error Fixing

Errors caught and fixed automatically

When your R Markdown fails to knit or your script throws an error, AiR detects it instantly. It reads the full error output, diagnoses the root cause, and proposes a fix - showing you exactly what changed with a color-coded diff.

  • Detects knit failures, console errors, and warnings
  • Reads full stack traces and render logs
  • Shows exact changes with accept/reject controls
  • Iterates if the first fix doesn't work
Learn more about Error Fixing
AiR - Viewer Pane
AiR
ChatDataTidyModel
regression.Rmd
Error in lm(price ~ carat + , data = diamonds)Fix
Console output:
Error in lm(price ~ carat + , data = diamonds):
unexpected ',' in formula
Execution halted
U
You
Fix it

Up and running in 60 seconds

Three commands. That's all it takes.

1

Install the R package

Install AiR from GitHub with a single command.

# Install AiR
devtools::install_github("useAiRStudio/air-package")
2

Authenticate

Run air_login() to link your account. A browser window opens for sign-in.

# Authenticate (opens browser)
AiR::air_login()

✓ Authentication complete
3

Enable auto-start

Run once - AiR will launch automatically every time you open RStudio.

# Auto-start on every RStudio session
AiR::air_install()

✓ AiR will now start automatically

Simple, transparent pricing

Start free. Upgrade when you need more.

Free

Get started with AiR at no cost.

$0 / forever
  • Limited monthly usage
  • Error fixing & diagnosis
  • Code editing with diffs
  • EDA reports (quick depth)
  • Smart context sync
Get Started
Popular

Pro

Full access for power users.

$24.99 / per month
  • 15x more monthly usage
  • All Free features
  • Deep EDA reports with statistical tests
  • Priority response times
  • Data Tidying: audit, transform, and export clean datasets
  • Model Builder with 15 model types, unsupervised mode & cross-validation
  • Diagnostic plots, model comparison & stepwise selection
Get Started
Privacy & Security

Your data stays yours

AiR is built with privacy at its core. We never train on your data, and the AI provider discards it immediately after processing.

Zero data training

Your code and data are never used to train AI models.

No data retention

Data is processed by the AI and immediately discarded. Nothing is stored on AI servers.

Encrypted in transit

All communication is encrypted via HTTPS/TLS.