Join the discussion
Write your take first — we'll ask for email only when you're ready to publish.
- Hacker News
- DSLs for AI doesn't really make sense- they were trained on existing graphics libraries and are pretty good at them.
Maybe a long term play would be putting this out and creating a "graphics bench" to entice the labs to overfit on your DSL but that seems like a lot of work
by siliconc0w - Why is this "for the AI era"? Isn't it kind of not for the AI era since an LLM can create a gnarly python or js chart of whatever you want in seconds?by woah
- It can write anything, but that's not a guarantee it can easily read and comprehend it later
- Nowadays you either create something with or for AI.
Like you created you own framework for wen apps before.
"For the AI era" is the necessary buzzword
by croes - 02 Jul 2026 04:28:21 UTC
Microsoft/Flint-Chart
https://github.com/microsoft/flint-chart
https://news.ycombinator.com/item?id=48756577
[ok]
08 Jul 2026 17:46:12 UTC
Show HN: Microsoft releases Flint, a visualization language for AI agents
https://microsoft.github.io/flint-chart/#/
https://news.ycombinator.com/item?id=48834924
[ok]
- 2,1,2
- Maybe they’re not selling this well. Maybe they’re so close to the AI research that this seems like an obviously good idea.
But there’s not one word of why this is good for LLMs, or how they tested/measured that.
My gut would tell me that a new solution put up against all the vega lite specs it’s already be trained on would be a hard thing to win.
- When does it get too abstract? What’s wrong with plotly? Or plotly express? How is yet another JSON spec era-anything?
- >When does it get too abstract?
When the abstraction detracts more value than it provides.
I think this is an interesting direction to explore. Plotly is fine. So is D3 and plenty of other solutions, but neither represent the ultimate pinnacle of charting.
- An earlier larger discussion 22 days ago on this https://news.ycombinator.com/item?id=48834924by thepoet
- What is wrong with "please make plot XYZ in plotly?"by Imanari
- If you're doing automated data visualization, "please make plot XYZ in plotly" needs a sandbox to execute the JS or Python code. With something like this you can avoid it, you just validate it and plot it. If the LLMs are better at generating Flint than Vega-Lite specs then I see how the project can be useful.
- A Microslop employee won’t get promoted for that.by xigoi
- It's cool...but is it needed? I'm thinking of Apache echarts and any of many other mature charting libraries. Kinda seems like reinventing the wheel; thing gets released, thing gets more and more new features bolted onto it, eventually someone offers another thing that's basically the original thing with some slightly different design and syntax opinions...
Switch backends to use their native strengths: ECharts for hierarchical sunbursts, Plotly for statistical and analytical traces, or Excel for editable charts embedded in a workbook.
Or just find a charting library that you like and actually get to know what it can do, vs mixing and matching presets from different libraries but never tweaking them.
by anigbrowl - So this is one interface that can render to multiple charting backends?
If AI is writing the "Flint", why not just have it write the backend code instead? I'm not sure why I would want pluggable charting backends.
I can see an argument for providing simpler APIs for LLMs, though, so that it can be more token efficient for example.
- Different charting backends support different kinds of charts. Being able to easily switch between "ECharts Sunburst" and "Vega-Lite faceted bar" - one of the examples on the project website - seems like a super useful ability.by awestroke
- I have tried using Flint vs asking the AI to generate a Vega lite spec directly, and in my opinion Flint was not as nice of a solution.
Flint is fine for doing predetermined chat types, with very low customization. But I found using an agent or sub agent to create the Vega spec directly allowed for a lot more flexibility, and ultimately that means higher quality visualizations (stuff like adding points for min and max on a timeseries, or adding a callout marker for a date where some event happened).
That being said, with Vega lite you have to validate your chart specs, provide specific guidance, and play whack a mole with Vega bugs/idiosyncrasies. So Flint is more reliable if you don’t want to dedicate a whole skill to making charts and want to get running quickly.
by data-ottawa - Even in the Era AI, GGPlot's API is still the best charting API. The name "Grammar of Graphics" isn't just marketing, they literally sought to write a god damn grammar to was capable of expressing all possible qualitative graphics.
They even wrote a book about how they went about it (not that it speaks to the quality of the API) https://link.springer.com/book/10.1007/0-387-28695-0
I actually stumbled upon this book when I was trying to look up how draftsmen (with pens and pencils on paper) did qualitative graphics as I found they had a lot of charm as opposed to modern charting libraries. It's something I noticed when looking through a bunch of historical RBA (Reserve bank of Australia) annual reports, the 1960-1980 charts had a lot of character, but then you go into the early 2000s and its a stale chart from excel.
Anyways ggplot doesn't really recapture the magic of those older charts, but it seems use quite a few of those as a baseline for how to communicate information. Like in figure 20.1 they talk about efforts to replicate older inforgraphics that showed Napoleon’s March on Russia, this graphic here (I think the example in the book is a bit nicer than the one in this blogpost IMO)
https://www.andrewheiss.com/blog/2017/08/10/exploring-minard...
On top of the charts just look nicer than anything you could produce with pyplot (and any API built on top of it) as pyplot seems to be have some really limited raster based rendering or something and the text handling is incredibly limited, I've never had this issue in ggplot.
I feel like most software engineers aren't exposed to because it exists in the R ecosystem which is more so data scientist, econometricians, statisticians and other quantitative data professions, but it definitely one of the nicer APIs and I wish more people in the node and python ecosystem copied their homework. I see vega's full name is something to do with grammars, but idk it's for the same reason.
by akst - I'm curious as to why that book is so god damn expensive.by darksim905
- Note that if you are affiliated with a University you probably can download a pdf of the book in the link for free.by nialse
- Well, for that hand-drawn charm if you're willing to use METAPOST there is:by WillAdams
- How would you quantify the magic of these older charts? What is it that makes them have charm and character?by porker
- Yeah Grammar of Graphics semantics are already very ideal for both humans and agents alike. Ggsql is a recent project that implements it as SQL user defined functions which is even more agent friendly
I dont see what this brings to the table
by nylonstrung - To be fair, the Vega-lite backend is json mostly based on GG ideas.
In this vein, I think I prefer ggsql that made it to HN recently [0].
[0]: https://ggsql.org/
by jiehong - Have you checked out plotnine?
https://raw.githubusercontent.com/rstudio/cheatsheets/main/p... it's made by the same people behind ggplot. Hadley Wickham worked for RStudio which also created plotnine.
plotnine is currently sponsored by Posit where Hadley works.
by abirch - > They even wrote a book about how they went about it (not that it speaks to the quality of the API)
Wilkinson’s Grammar of Graphics inspired ggplot. The textbook doesn’t even mention ggplot.
Shameless plug for my own GoG inspired DSL, Algraf:
https://williamcotton.github.io/algraf/demos
There’s a Minard plot in the demos!