You can integrate sqlite-vec into SQLAlchemy models without too much work.

github.com/ploomber/doc...

I have a new use for AI: create image backgrounds

We just released an update to our Shiny AI editor! Code generations are now much more accurate. Try it out! #rstats

editor.ploomber.io

Thanks! We're working on improving bslib support! For now, you can try attaching the image and add "use bslib" to the prompt.

It might yield apps with missing `library(bslib)` calls but that should be fixed in our next release. Feel free to share your feedback!

We've made a lot of changes last week to improve the quality of Shiny app generations. I'm pretty excited about this! We'll release this in the next few days! #rstats

Nice.

Claude 3.5 Sonnet + some RAG that uses ggplot's documentation! We'll ship an initial version next week but we'll keep working in improving generation quality.

a cartoon character says witchcraft in front of a curtain ALT: a cartoon character says witchcraft in front of a curtain

yes!

Step 1: Drop an image

That's it. Just drop an image to generate a fully functional Shiny app. #rstats

(coming up next week to editor.ploomber.io)

Code an app from an image coming up soon to our editor! Here's a #Streamlit demo:

Some stuff we're shipping soon: Microsoft Entra (Azure Active Directory) authentication for Streamlit and Shiny.

You can select Shiny, type the description, and then click on "Create Shiny app." That's how I generated the time series visualization.

Ploomber AI Editor | timeseries-analysis-e649 Interactive Shiny app: create an app that creates time series data for multiple categories. use ggplot for visualization and add markers and labels for important points, add...

You should be able to use it right away but some features will prompt you to create an account (like saving an app or to increase rate limits): editor.ploomber.io - I'm pending on creating an onboarding flow to explain all the features so stay tuned! I'd highly appreciate any feedback!

It implemented some logic to detect local maxima/minima and then samples them. You can see the code here: editor.ploomber.io/editor/times...

Wow. Microsoft just released a tool to convert office documents to Markdown. This unlocks a lot of enterprise use cases for LLMs.

What do you think? This was generated with a single prompt without any further edits.

editor.ploomber.io/app/timeseri...

Makes sense. There are tons of “single player” use cases.

SQLite in the browser is amazing! How do such apps keep the state of user interactions? eg user A posted an updated, and user B should see it.

Interactive Table in Streamlit with AgGrid Build interactive table from Dataframe with AgGrid

Streamlit: https://ploomber.io/blog/streamlit-aggrid/
Shiny: https://ploomber.io/blog/shiny-bslib/

WASM is enabling a new generation of high-performance web applications that were previously impossible. The browser is becoming an actual universal runtime.

Both run entirely in your browser through WASM - no R or Python servers are required. Every interactive table, filter, and visualization executes directly in your browser's sandbox, showcasing WASM's ability to bring powerful data science tools to the client side.

Imagine running Python or R directly in your browser at near-native speeds. There are no servers, no API calls, just pure client-side computation.

We've published a technical demonstration featuring two popular data science frameworks: R's bslib for Shiny and Python's Streamlit with AG Grid.

WebAssembly (WASM) is changing how we think about web applications. For decades, browsers have been JavaScript's exclusive domain - if you wanted to run code in the client, JS was your only option. WASM breaks these chains.

Faster debugging in editor.ploomber.io!

1. The editor detects issues in your app
2. See the error in the terminal
3. Clicking on "Help" passes the error to the AI chat

Works with Shiny and Streamlit! #rstats

I love this new stuff in editor.ploomber.io! Tag files in the code generation prompt to pass them to the model.

I have a Shiny app (works with Streamlit) with two files: UI and logic. I move to the UI file and ask the model to update the UI based on the second file #rstats

I loaded ggplot's docs into our Shiny AI editor to improve code generation. I generated an advanced (?) histogram and it worked pretty well.

What kind of visualizations should I try? I want to test this with more difficult viz ideas. #rstats

We just shipped a new feature to our Shiny/Streamlit editor: tag file with @filename to ask questions in the chat. Great for debugging! #rstats

https://buff.ly/4f9EFxI

I just updated my Bluesky sentiment analysis app. Using bslib makes it much prettier!

The sentiment for "Manhattan" has shifted negatively compared to two weeks ago (we all know why).

Try the app here: editor.ploomber.io/app/d2a65731

#rstats

I updated my Bluesky sentiment analysis app to compare terms across time and added a cache via memoise. Next, redo the UI with bslib. #rstats

You can try the app here: editor.ploomber.io/app/d2a65731

Got a sentiment analysis app (using Bluesky's API) running entirely in the browser with Shinylive!

The main limitation is that curl (a dependency for most HTTP R libraries) doesn't work in Shinylive, but fromJSON does!

editor.ploomber.io/editor/d2a65...

#rstats

Ploomber AI Editor Create custom Streamlit and Shiny R apps effortlessly with AI assistance. Design, code, and deploy data apps in minutes.

I'm using this: editor.ploomber.io?framework=sh...

(which I'm building with my team, so feel free to share any feedback!)