Coding up a storm of bits. Building AI products for @zardotapp. The Ruby AI Podcast co-host. @RubyRogues panelist. aka @codenamev@ruby.social
I am here.
- Tweets 786
- Following 472
- Followers 320
- Likes 4,320
Pinned Tweet
Joe Leo and I have officially launched a very special first episode of The Ruby AI Podcast: "Beyond Chat: Phoenix Tests, Ruby Agents & the AI Tipping Point"
In the pilot episode, we chat about shipping real AI products in Ruby, how Phoenix got started, and why Ruby syntax actually makes this weird new world fun to build in.
Hope you'll come along for the ride.
(And send us your favorite AI coding mishaps—we might just feature them in a future episode!)
therubyaipodcast.com/
1
3
8
1,075
Good evening Rails World! Ruby Roles is live - find the best matching career opportunities for you.
Hope you had a great time at the conference and are out having some fun! #railsworld
rubyrole.com/
1
8
47
2,571
Hey @railsworld! As usual I've got FOMO, and as usual I'm releasing something into the #ruby wild!
Introducing: ruby-laya
github.com/codenamev/ruby-la…
It's a #Ruby port of the open-weight System 1 decision engine Laya. Now you can get all your agent questions answered at blazing speed (33ms!) on your CPU at zero cost 🎉
5
5
46
4,235
@brainFnCl has some great examples of where it shines
This is NOT Jev.
Open source. Runs on your laptop. Decides in ~27 ms, about 200× faster than waiting on a hosted LLM.
Here it is playing Tetris by itself 👇
brainfunctioncollapse.com/la…
1
361
Valentino Stoll retweeted
Jev and the System One Model: RLCD, intelligence/$, reliable AI, & the end of chat-first AI latent.space/p/jev
@typesafeai CEO @CompleteSkeptic explains why AI can solve extraordinarily hard problems yet still fail to automate basic work, why Jev is built for reliable decisions inside software instead of chat, why TypeSafe rejects public benchmarks and refusals at the API layer, why data and the right task matter more than brute-force compute, how System One Models could reshape coding agents and software, and why even with $1 billion he wouldn’t pre-train a model from scratch.
Jev pod tomorrow
subscribe on apple /youtube @latentspacepod
thanks to @allenpark and Ke for making this happen!
61
47
434
381,978
So sick. RIP eval harnesses.
Just landed in RubyLLM's main:
`RubyLLM::Judge` with support to @typesafeai's Jev
rubyllm.com/next/judgments/
3
8
1,090
Valentino Stoll retweeted
Got this comment via text message directly from a longtime friend who's not on X and it was too good not to share:
"Insightful and aligns with my current experience too. I finally got over my trust issues once Fable came out. I review plans now to ensure correct architecture not code anymore. I’ll occasionally glance at PR code out of curiosity, but mostly I just look at the summary, merge and move on. At work we just did a long overdue redesign of our user flow that looks like it has increased revenue 30-40% which would have taken many man months to do before, and something we probably would not have even attempted to tackle with our small team of 3, but we knocked it out in two weeks and business is reaping the benefits. It’s time to be ambitious!
My focus has become how do I make this business as successful as possible so that I have something to do in the future."
Your job may be surviving on inertia. Your resentment won’t extend the runway. What do you actually want to do with the intelligence now at your disposal?
9
10
132
16,600
Valentino Stoll retweeted
Introducing feelings gem
A decision model can answer a typed question about a value with a calibrated probability. Probably showed what code reads like when an if can ask one of those questions directly instead of hand-rolling a prompt. feelings implements that kind of construct in beautiful, idiomatic Ruby.
github.com/obie/feelings
I've seen people describe Jev as an "AI if statement". But what if it actually WAS an if statement?
Introducing Probably: a programming language powered by Jev: probably-lang.southpolesteve…
Jev baked into the language. “feels” asks a question. “match” routes between descriptions. “while” keeps going until something stops feeling true.
This is obviously a toy, but it's fun to think about what something like Jev unlocks. Jev makes the decisions, an LLM does the writing, and a little program ties it together.
4
34
4,932
Valentino Stoll retweeted
Hey rubyists, here's the best way to play with these new decision models like JEV from @typesafeai
Full-featured, and future-proof. Introducing github.com/obie/ruby_decisio…
3
21
119
5,642
This.
Mark my words: @typesafeai JEV is not only the new hottness, it's a genuine phenomenon as a whole new wave of normie programmers learn how to plug fast, cheap intelligence into their otherwise normal codebases.
Feels like Xmas 2025 all over again. I predict that all the major frontier labs will have JEV-style offerings within the next few months. I will not be surprised one bit if decision models end up being bigger than LLMs in the long run.
If you think I'm being hyperbolic, consider that hacking LLMs into production codebases has literally been my focus for over 3 years now and I wrote the definitive book on the subject: Patterns of Application Development Using AI
We are definitely in a singularity, because nobody I know saw this coming.
1
193
Hey @paolino do you have any magic incarnations for your electron app re-writes? Would love to recreate one for @Mattermost 🥺
1
222
Human layer has always been ahead of its time 😊 this is what I thought of when I saw Jev too!
jev is the best excuse you could possibly have to go re-read 12 factor agents. Tool calling itself can be decomposed into classify+action,
if you learn to design ai programs as pipelines that switch breathlessly between classification, structuring data, deterministic code, AND small agent-shaped append-chat loops, then jev is a WONDERFUL building block
hlyr.dev/12fa
107
Developers never wanted text. This makes a ton of sense and looking for to the explosion in speed optimizing for the right output shapes 🤞 I see it
The gains aren’t free: Jev can't generate text
Comparing Jev vs LLMs side-by-side makes the trade-off clear
Fun fact: replacing sequential computation with parallel is the same way Transformers leapfrogged RNNs
3
172
I absolutely love this idea! TL;DR bundle knowledge into your gems so everyone's agents can take advantage of it
Shipping skills could soon mean as much as READMEs. As AI writes more, libraries that teach agents to correctly use their APIs will win. Rails Hyperdrive lets gems distribute agent guidance, and getting started just takes Markdown files and a gemspec metadata entry.
evilmartians.com/chronicles/…
1
4
450
Valentino Stoll retweeted
Voice, music, image, and video just went live in the ElevenLabs MCP.
To celebrate, we're giving 111 people 11k free credits.
To enter:
1. Comment "MCP" below.
2. Retweet the post.
In 24 hours we'll pick 111 who do both and DM the credits.
Introducing voice, music, image, and video generation in the ElevenLabs MCP.
Generate speech, transcripts, dubs, music, sound effects, images, and video from the assistant you already work in.
140
128
59
2,735
Damn, spotifast.rocks was such an incredibly smooth install and experience 👏 Never underestimate a Rubyists' freetime joy, amirite? Thank you for this @paolino
3
884
Valentino Stoll retweeted
RubyLLM 2.0 is the biggest update RubyLLM has ever seen.
Here's everything that's new.
Also, follow @berlinrb_org for more!
piped.video/6FAexWLxu6w?si=cnOz…
12
46
3,201
Valentino Stoll retweeted
Just recently I came across a really cool blog post from Evil Martians about TutorialKit.rb.
I decided to experiment and used the learnings from it to create a set of interactive tutorials designed to help you learn the basics of RubyLLM in about a lunch hour.
I figured I would share it here!
Also - I am especially interested in feedback, whether that is on the tutorials themselves, the pacing, clarity, or anything that felt confusing or could be improved.
learnrubyllm.com/
3
5
14
2,920
Valentino Stoll retweeted
Rails + AI are a fantastic pair. What's the best model for your needs?
We partnered with @rails to benchmark different models and answer that question.
The study happened in two stages:
- Stage one: Learning how well models understood Rails and performed with atomic tasks.
- Stage two: Studying how efficient these models are when solving feature tickets, like a human developer would do.
Here's a summary of the findings. Detailed information below.🧵
1
11
32
1,470
A new generation is waking up to the wonders of software with agentic coding, and when they look inside what they've built, I would love for them to see the joy of Ruby rubyai.beehiiv.com/p/one-dec…
6
1
18
2,927
"A Rails app whose screens are composed while you talk."
~ github.com/roboruby/poetry_i…
This is one Bad Mamajama! 😂 Thank you @mattsolt 🙇
6
331