Introducing AI in RStudio
An entirely new way to work with data
Over the last few years, large language models have exerted a massive gravitational pull on knowledge work. Software engineering in particular has changed almost overnight. As the creators of open-source tools used by millions of people every week to analyze data, we’ve watched this shift with a heavy dose of skepticism.
Writing a web application is fundamentally different from analyzing scientific data. It’s relatively straightforward to verify if pure code works and does what the user intended. If an LLM writes a bad piece of boilerplate UI code, a button in your application might not work.
The high stakes of data analysis
But the fundamentals of sound data analysis are different. If an LLM hallucinates a data transformation, misinterprets a plot, or suggests the wrong statistical method, the problem may be difficult to detect, and the consequences may be severe. In fields like public health or finance, a wrong answer delivered confidently can have dangerous consequences which harm real people.
Because of these risks, we were initially concerned about the integration of AI into data science workflows. Part of that caution manifested in making tools for very specific purposes. We developed and shipped tools like Positron Assistant and Databot to learn exactly how AI can safely assist data practitioners in constrained environments like exploratory data analysis.
Alongside our own development, the timeline of model advancements accelerated our excitement. As 2025 marched on, the increase in performance of state-of-the-art LLMs on data science tasks was noticeable in our internal experiments. By December 2025, it was undeniable. We shifted from being happy when LLM-assisted data science worked to being surprised when it didn’t.
The utility of AI for data analysis is too great to ignore
Our conclusion from many months of building, testing, and shipping is crystal clear.
Despite what we initially expected or feared, the usefulness of a capable language model and well-designed agent harness when applied to data science and analysis is, frankly, staggering, and cannot be ignored.
When a new technology arrives that fundamentally alters how fast we can move, society rarely rejects it. The invention of the automobile introduced entirely new, life-threatening risks to daily life. Yet, the sheer utility of motorized transport was so valuable that we did not ban cars. We paved roads, built traffic lights, and designed seatbelts and airbags. We engineered systems to mitigate the risk, and we collectively decided to accept some risk in return for the benefit, rather than abandon progress.
The same is true for AI in data science. There are risks, but because these tools are so useful, there are also possible harms in not moving forward with AI. RStudio users do important work, and we believe they should have access to the most useful tools we can provide to advance that work. The risks are real, the downsides are significant, but the utility is simply too great to ignore. We can’t afford to sit on the sidelines while the tools for doing our work fundamentally change around us. We want you to have the opportunity to participate in this transformative new way to work with data, too.
Today, we’re announcing that AI in RStudio is here. We’ve embedded a specialized agent, Posit Assistant, directly into RStudio, transforming the IDE into a powerful, context-aware collaborator for data analysis.
The problem with context
Using AI for data analysis typically means working in two disconnected worlds. You encounter a complex data analysis problem, copy your code, tab over to a chatbot in your browser, paste in your code, and hope for the best.
But the chatbot can’t see your data. It doesn’t know what packages you have loaded, what variables are in your environment, or what your data frame actually looks like. It’s guessing. And when language models guess in data science, they can hallucinate or make mistakes.
As Hadley Wickham, Posit’s Chief Scientist and the creator of the Tidyverse, explains:
“Bouncing between a chatbot and RStudio feels like the old days of copying and pasting plots into a Word document. And because the chatbot doesn’t know what’s in your R session, it often gives the wrong answer. Posit AI closes the loop by embedding the chatbot in RStudio, letting it see what data and packages you have loaded and run code on your behalf.”
Introducing Posit Assistant and Next Edit Suggestions
You can use AI in RStudio in two new ways.
Posit Assistant represents the evolution of everything we learned from Positron Assistant and Databot. It’s a conversational agent that lives in your RStudio workspace. You can ask it to visualize patterns, reshape datasets, debug errors, build Shiny apps, explain code, and much more. Because it can read your live R session context, it provides highly precise answers based on your actual data, not generic assumptions.
Next Edit Suggestions (NES) operates in the background, offering targeted code completions as you type. It speeds up your coding process by surfacing syntax for joining tables or updating parameters, based on what you are actively working on, without breaking your concentration.
Security and Zero Data Retention
We know that for organizations and researchers handling sensitive data, privacy is a strict requirement. Sending proprietary data to a public chatbot is a severe security risk.
We’ve established a Zero Data Retention agreement with our language provider, Anthropic. When you send a prompt or code snippet through Posit AI, it is used exclusively to generate a response and is then immediately deleted. Your code, data, and text are never stored on their servers and are never used to train future models.
But what about the data Posit collects? When you create your account, you can opt-in to sharing trace data—basic logs of how the tools are used—with Posit to help us improve Posit Assistant and Next Edit Suggestions. By default, you’ll be opted out.
Pricing
For over fifteen years, we’ve built RStudio as a free, open-source IDE for data science, and RStudio itself will continue to remain free and open-source. Posit’s mission is to create free and open-source software for data science, scientific research, and technical communication. Revenue from Posit’s commercial products is what funds this open-source work and makes our business sustainable, and Posit AI is one of those commercial products. As a managed service, it also allows us to tune the experience for high-quality models and maintain reliability for users.
Posit Assistant and Next Edit Suggestions are accessed in RStudio exclusively through our new subscription service, Posit AI, starting at $20 per month.
Here’s exactly how the plan works:
- Unlimited Next Edit Suggestions: Your subscription includes unlimited use of NES, Posit AI’s background code completion tool. You don’t have to worry about hitting a limit while writing code.
- Posit Assistant: Posit Assistant requires significantly more resources to operate. Your $20 monthly plan includes an allowance of credits specifically for Posit Assistant. For many users, this covers a full month of typical use.
- Automatic Recharge: If you’re working on an intensive project and reach your Posit Assistant usage cap, your work doesn’t have to stop. You can opt into automatic recharge to add additional credits exactly when you need them. You control your budget, only pay for what you actually use, and are able to rollover unused additional credits into the next month.
Posit AI is, and will continue to be, an optional service which powers Posit Assistant and Next Edit Suggestions. Using RStudio without AI will always remain free. You don’t need to pay anything to continue using RStudio without AI.
Posit Assistant and NES in RStudio Desktop and RStudio Server require the managed Posit AI service and do not support custom model providers via ‘Bring Your Own Key’ (BYOK). This helps us ensure an accurate, high-performance experience powered by frontier-grade models that are most effective for the complex reasoning required in data science.
We recognize, however, that enterprise requirements differ significantly from those of individual users. To address the specific needs of organizations regarding data boundaries and the use of existing, internally approved model providers, we are prioritizing BYOK and custom provider support specifically for Posit Workbench in 2026.
While Posit AI will remain a managed service to guarantee a quality experience and operational sustainability, this dual approach allows us to deliver best-in-class intelligence to individuals while giving enterprises the architectural control they require.
Try it for free
We built RStudio to give you a place to think clearly about data. Posit Assistant is the next step in protecting that focus.
It’s available today in the daily build of RStudio. If you want to activate this new AI functionality in RStudio, you’ll be prompted to create a Posit AI account. There are no complicated settings to configure. Just download the latest daily build and set up your account. When you start a free trial, we give you a grant of one-time credits upfront. There’s no credit card required.
If you’re wondering if the time to start using AI for data analysis is now, we encourage you to try it. Open RStudio, load the messiest dataset you have, and see what it feels like to have an assistant that actually understands your environment.
