Simon P. Couch
LightNews LightNews
banner
simonpcouch.com
Simon P. Couch
@simonpcouch.com
3.5K followers 200 following 230 posts
he/him - writing statistical software at Posit, PBC (née RStudio)🥑

simonpcouch.com, @simonpcouch elsewhere
Posts Replies Media Videos
simonpcouch.com
infer 1.1.0, a package implementing an expressive grammar for statistical inference, is on CRAN! In a change originally motivated by @allendowney.bsky.social's posit::conf(2024) keynote, the package now supports arbitrary test statistics.

Read more: github.com/tidymodels/i...
A hexagonal logo for "infer" featuring a green Douglas Fir tree centered on a mountain landscape background with blue-gray peaks and white snow-capped ridges. The word "infer" appears in a modern, lowercase black font at the bottom of the hexagon, which has a dark blue border.
December 18, 2025 at 2:27 PM
simonpcouch.com
GPT 5.2 is now included in our R code generation eval! The -Pro version is slightly SoTA at a substantially higher price point than similar performers.

Read more: skaltman-model-eval-app.share.connect.posit.cloud
A scatter plot comparing AI models by R coding accuracy (percent correct) versus total cost in USD, with points colored by provider (Anthropic in purple, Google in yellow, OpenAI in blue). Claude Opus 4.5 and GPT-5.2 Pro achieve the highest accuracy (~70%), but Claude Opus 4.5 does so at roughly one-sixth the cost of GPT-5.2 Pro.
December 12, 2025 at 6:20 PM
simonpcouch.com
A new release of chores is now on #rstats CRAN!

There are now some LLMs tiny enough to run on a laptop that are capable of powering chores helpers. Here's a real-time demo with a model running on my Macbook.

Read more: www.simonpcouch.com/blog/2025-12...
December 10, 2025 at 3:25 PM
simonpcouch.com
A new release of odbc, a package that allows for connecting to various databases, is on #rstats CRAN! Just a few bug fixes in this release.

github.com/r-dbi/odbc/r...
A diagram containing four boxes with arrows linking each pointing left to right. The boxes read, in order, "R interface," "driver manager," "ODBC driver," and "DBMS." The left-most box, R interface, contains three smaller components, labeled "dbplyr," "DBI," and "odbc."
December 8, 2025 at 2:47 PM
simonpcouch.com
I get the appeal of local LLMs—free, private, no big tech. That said, they're not yet ready to power coding agents like Positron Assistant or Databot.

www.simonpcouch.com/blog/2025-12...
Scatter plot revealing the cost-performance tradeoff across nine LLMs. Claude models occupy the upper-middle, with perfect accuracy and moderate cost. Gemini models tend to be cheaper but less performant. OpenAI models are more variable. The local (and budget Flash and Mini) models cluster in the lower-left, with minimal cost but zero performance. The models closest to the 'top left', being cheap but powerful, are Claude Haiku 4.5 and Gemini Pro 2.5.
December 4, 2025 at 2:27 PM
simonpcouch.com
It's Spotify Wrapped season🎁 Continuing the tradition of analyzing my music listening data with #rstats tidyverse, this time with Databot: www.simonpcouch.com/blog/2025-12...
A screenshot of the beginning of a conversation with databot, where I say 'Read blog/2025-12-03-wrapped-databot/Library.xml and turn it into a tibble' and the agent begins working.
December 3, 2025 at 9:57 PM
simonpcouch.com
A new release of vitals, a package for LLM evaluation in #rstats, is now on CRAN!🎄 This release includes all sorts of quality-of-life improvements; image support, precise latency measurement, better error messages, and more comprehensive logging.

github.com/tidyverse/vi...
December 2, 2025 at 2:17 PM
simonpcouch.com
Do you use Positron?😉
November 20, 2025 at 4:20 PM
simonpcouch.com
Gemini 3 is out! I've been poking at it with `side::kick()` and am impressed so far.

In case you want to connect to it with the #rstats ellmer package, I wrote some quick notes in a Gist: gist.github.com/simonpcouch/...
November 18, 2025 at 9:54 PM
simonpcouch.com
A core capability for data science agents is making plots and learning from them. While developing Databot and Positron Assistant, though, we've seen that LLMs tend to ignore plotted trends when they're counterintuitive.

From @sara-altman.bsky.social and I: posit.co/blog/introdu...
A stacked bar chart illustrates model hallucination on built-in R datasets like mtcars, showing the proportion of correct versus incorrect responses from three language models: GPT-5, Gemini Pro 2.5, and Claude Sonnet 4.5. All three models demonstrate similarly poor performance, with approximately 90-95% incorrect responses (shown in coral/orange) and only 5-10% correct responses (shown in blue), indicating a tendency to report expected characteristics rather than accurately describing the plotted data.
November 18, 2025 at 3:46 PM
simonpcouch.com
side::kick() is model-agnostic. When you first launch the app, a setup flow will detect whether any common providers are already configured to help you get started.

#rstats
November 7, 2025 at 12:48 PM
simonpcouch.com
I'm excited to share side::kick(), an experimental open-source coding agent for RStudio built entirely in R. It can interact with your files, communicate with your active #rstats session, and run code.

Check it out: github.com/simonpcouch/...
November 5, 2025 at 3:57 PM
simonpcouch.com
I'll be keynoting at R/Pharma a week from today! The conference is free and virtual. I'll be focused on the mundane use cases of LLMs for wrangling data with #rstats, and the content should feel applicable for folks outside of pharma—come through. :)

Register: events.zoom.us/ev/Ai-geyS63...
A title slide reading "Practical AI for data science" and my name and title.
October 29, 2025 at 7:32 PM
simonpcouch.com
I sincerely apologize for the mcptools package not having a cute lil feller on its hex sticker in its first release. This will be corrected in the next #rstats CRAN release!!!

posit-dev.github.io/mcptools/
A before-and-after comparison of hex logo designs for "mcptools". The "Before" panel features a serene landscape with a wooden bridge over water, surrounded by trees, flowers, and mountains in muted tones. The "After" panel presents a vibrant, cartoon-style jungle scene with a cheerful monkey hanging from a rope bridge above a waterfall, holding a wooden sign that reads "mcptools", with colorful flowers and lush greenery throughout.
October 28, 2025 at 6:25 PM
simonpcouch.com
ICYMI, @sara-altman.bsky.social and I have been writing a biweekly newsletter on AI and open source data science on the @posit.co blog!

A bit about how that came to be on my #rstats blog: www.simonpcouch.com/blog/2025-10...
A screenshot of the Posit Blog homepage showing statistics (936+ posts, 22+ categories, 386+ tags) and two featured blog post cards. Both posts are AI Newsletter roundups from September 2025 by Sara Altman and myself, featuring AI-related R package hexes in their hero images.
October 8, 2025 at 1:16 PM
simonpcouch.com
Just dropped a WHOLE BUNCH of these somewhere👀👀👀

#positconf2025
A picture of a stack of hex stickers for the stacks package. The stickers have a tan and orange peaches & cream theme rather than the usual blueberry.
September 18, 2025 at 3:21 PM
simonpcouch.com
Just wrapped up a talk at #positconf2025 about LLM evaluation with R! Such a joy to hang out with #rstats folks in person and hear about what others are working on.

Resources and slides: github.com/simonpcouch/...
A chat interface showing a conversation starting with a user saying "Are there any talks about evals" and a response "Yes, Simon Couch just gave one, and he hopes you enjoyed it." There's a teddy bear mascot wearing a teal shirt and cap in the bottom right corner of the chat window. A picture from the side of a conference session room, with myself standing in front.
September 17, 2025 at 7:08 PM
simonpcouch.com
@sara-altman.bsky.social and I just shared the second release of the Posit AI Newsletter! In this edition, a landmark settlement on LLM training data and copyright, the emergence of agentic browsers, and more.

Read and subscribe on the Posit Blog: posit.co/blog/2025-09...
A graphic displaying, on the left, three robot icons, and on the left, a selection of LLM-related R package hex stickers.
September 12, 2025 at 4:10 PM
simonpcouch.com
So stoked to share about the first release of the Posit AI Newsletter!

@sara-altman.bsky.social and I have been working on a pilot of this newsletter internally for a few months now—an opinionated and pragmatic snapshot of AI news released every two weeks.

First edition: posit.co/blog/2025-08...
A graphic displaying, on the left, three robot icons, and on the left, a selection of LLM-related R package hex stickers.
August 29, 2025 at 3:01 PM
simonpcouch.com
In working on an eval for an experimental tidymodels AI assistant, I realized that today's frontier LLMs know much more about #rstats tidymodels than I thought.

www.simonpcouch.com/blog/2025-08...
A ggplot2 bar plot. On the x axis are two LLMs, Claude Sonnet 4 and Gemini Pro 2.5. Both bars are dodged by three configurations: Predictive, Databot, and `run_r_code()` tool. On the y axis is relative performance, ranging from 0 (baseline model) to 100 (best human modelers). The bar heights for Claude Sonnet 4 are all about the same (around 80) and, for Gemini 2.5 Pro, the height for Predictive, an app designed specifically as a tidymodels assistant, is much lower (around 20).
August 26, 2025 at 6:42 PM
simonpcouch.com
The limited run of the #rstats stacks hex for this year's posit::conf is confirmed🥞🥞🥞 I'll have a batch of peaches & cream stickers, come say hi to pick one up!

posit.co/conference/
Variants of the hex logo for the stacks package from different posit::conf years, each featuring a stack of pancakes with different toppings and color schemes. The 2025 version (top left) has orange colors with peach slices, 2024 (top right) uses dark blue/purple with blackberries, 2023 (bottom left) features red/pink with raspberries, and the original (bottom right) displays teal/blue with blueberries.
July 25, 2025 at 2:38 PM
simonpcouch.com
Drawing comparisons to Deepseek, a little-known company dropped an open-weights LLM Kimi K2 on Friday with some very impressive benchmarks. I tried it out and the vibes were bad.

www.simonpcouch.com/blog/2025-07...
A horizontal bar chart comparing various AI models' performance on R coding tasks. The chart shows percentages of correct (blue), partially correct (beige), and incorrect (orange) answers. Models listed from top to bottom are Kimi K2, Claude 4 Opus, Claude 4 Sonnet, GPT 4.1, GPT 4.1-mini, and Gemini 2.5 Flash. Claude 4 Opus shows the highest proportion of correct answers at approximately 60%, while Kimi K2 sits around 36%.
July 14, 2025 at 4:04 PM
simonpcouch.com
I've been getting a lot more out of Claude Code after hooking it up to tools that let it read #rstats pkg docs. I just type "read about some::topic" before tasking Claude with things that are unlikely to be in its training data, and the model will pull up whatever help files it needs.
June 11, 2025 at 2:15 PM
simonpcouch.com
Introducing acquaint, an R package that turns your R sessions into a Model Context Protocol (MCP) server. This allows MCP-enabled tools like Claude Desktop and Claude Code to run #rstats code _in your active R sessions_ to explore objects, read documentation, etc.

posit-dev.github.io/acquaint/
May 28, 2025 at 3:43 PM
simonpcouch.com
New on my blog: the Claude 4 models are here! I evaluate the new releases of Sonnet and Opus against Claude 3.7 Sonnet and o4-mini on a dataset of challenging #rstats coding problems.

www.simonpcouch.com/blog/2025-05...
A horizontal bar chart comparing various AI models' performance on R coding tasks. The chart shows percentages of correct (blue), partially correct (beige), and incorrect (orange) answers. Claude 4 Opus has the highest proportion of correct answers, followed by o4-mini, Claude 4 Sonnet, and Claude 3.7 Sonnet.
May 27, 2025 at 3:48 PM