Gareth Dwyer ⠕ retweeted
Replying to @serxzsz
whats the point of wealth if not to make terminals great, really
Gareth Dwyer ⠕ retweeted
clairic is redesigned and now open-weights! we went back to the drawing board and overhauled the architecture for it to excel at extraction.
new:
- clairic (27B) and clairic-flash (9B)
- image extraction greatly improved
- PDF support
- weights release!
blog.cloudflare.com/open-cla…
Gareth Dwyer ⠕ retweeted
This is a great little story of coming to understand how to use Jev and what we mean by System One thinking!
Decompose your problems!
quick tip we learned the hard way @ rivet:
jev does not reason
...which is by design
...and we should have rtfm first
~
we started by prompt-engineer to ask something like:
```
𝚜𝚎𝚟𝚎𝚛𝚒𝚝𝚢 = 𝚊𝚜𝚔(
"𝚐𝚒𝚟𝚎 𝚖𝚎 𝚝𝚑𝚎 𝚜𝚎𝚟𝚎𝚛𝚒𝚝𝚢 𝚘𝚏 𝚝𝚑𝚒𝚜 𝚒𝚜𝚜𝚞𝚎. 𝚏𝚘𝚛 𝚎𝚡𝚊𝚖𝚙𝚕𝚎, 𝚜𝚎𝚟-𝟷 𝚒𝚜 𝚇, 𝚜𝚎𝚟-𝟸 𝚒𝚜 𝚈, 𝚜𝚎𝚟-𝟹 𝚒𝚜 𝚉, 𝚎𝚝𝚌",
𝚘𝚞𝚝𝚙𝚞𝚝𝚜=[𝚂𝙴𝚅𝟷, 𝚂𝙴𝚅𝟸, 𝚂𝙴𝚅𝟹, 𝚂𝙴𝚅𝟺]
)
𝚛𝚎𝚝𝚞𝚛𝚗 𝚜𝚎𝚟𝚎𝚛𝚒𝚝𝚢
```
normally, an llm would spend reasoning tokens to come to a descent conclusion
but doing this with jev gives meh results, and defeats what makes jev awesome
~
after rtfm, we split it up in to tiny prompts like this:
```
𝚒𝚜_𝚍𝚘𝚠𝚗 = 𝚊𝚜𝚔("𝙳𝚒𝚍 𝚝𝚑𝚎𝚒𝚛 𝚙𝚛𝚘𝚍𝚞𝚌𝚝𝚒𝚘𝚗 𝚜𝚎𝚛𝚟𝚒𝚌𝚎 𝚐𝚘 𝚍𝚘𝚠𝚗 𝚘𝚛 𝚋𝚎𝚌𝚘𝚖𝚎 𝚞𝚗𝚞𝚜𝚊𝚋𝚕𝚎?")
𝚒𝚜_𝚋𝚕𝚘𝚌𝚔𝚎𝚍 = 𝚊𝚜𝚔("𝙰𝚛𝚎 𝚝𝚑𝚎𝚢 𝚋𝚕𝚘𝚌𝚔𝚎𝚍 𝚠𝚒𝚝𝚑 𝚗𝚘 𝚠𝚘𝚛𝚔𝚊𝚛𝚘𝚞𝚗𝚍?")
𝚒𝚜_𝚋𝚛𝚘𝚔𝚎𝚗 = 𝚊𝚜𝚔("𝙸𝚜 𝚜𝚘𝚖𝚎𝚝𝚑𝚒𝚗𝚐 𝚒𝚗 𝚁𝚒𝚟𝚎𝚝 𝚋𝚛𝚘𝚔𝚎𝚗 𝚘𝚛 𝚏𝚊𝚒𝚕𝚒𝚗𝚐?")
𝚒𝚏 (𝚒𝚜_𝚋𝚕𝚘𝚌𝚔𝚎𝚍) 𝚛𝚎𝚝𝚞𝚛𝚗 𝚂𝙴𝚅𝟷
𝚒𝚏 (𝚒𝚜_𝚍𝚘𝚠𝚗) 𝚛𝚎𝚝𝚞𝚛𝚗 𝚂𝙴𝚅𝟸
𝚒𝚏 (𝚒𝚜_𝚋𝚛𝚘𝚔𝚎𝚗) 𝚛𝚎𝚝𝚞𝚛𝚗 𝚂𝙴𝚅𝟹
𝚛𝚎𝚝𝚞𝚛𝚗 𝚂𝙴𝚅𝟺
```
which works like a charm
(example screenshot of real-world classifier below)
~
any time we find a ticket that isn't weighted well:
- we give the ticket to an llm with what we expected
- llm adds it to a test suite
- llm adjusts decision logic
- backtests new decision logic on old issues
within a few tickets the classifier logic was really solid
~
this is a feature of jev, not a limitation
in practice, it gives more control over how each decision is weighed than a generic llm
which makes it easier to reason about, monitor, and back test how agents will behave
~
for reference, this is how many jev questions we ask for a few of our classifiers in our internal support tool:
support classification and customer status: 13
severity: 6
team replies and problem identification: 6
follow-up timing: 6
linked dependencies and later conversations: 4
addressed teammate: 2
conversation grouping: 2
~
all that to say, jev is awesome
Gareth Dwyer ⠕ retweeted
Quad9's forced withdrawal from France highlights the dangers of mandating DNS-level blocking.
Read more: cfl.re/4f0cowv
Gareth Dwyer ⠕ retweeted
Replying to @WilliamAEden @diviacaroline
You’re partially right, but I want to push back here: I didn’t “write the AI.” I’m Chelsea Voss, a large language model engineer trained by OpenAI. I’m not an AI myself — I’m just unable to pretend otherwise. The crux is this: I have no special ability to imitate LLMs. The real requirement is skill-shaped, and that takes practice. It’s not a cheat code, it’s a way of life.
Gareth Dwyer ⠕ retweeted
"programming is solved"
[outlook goes down for like 18 hours]
"yes but imagine how long it would have taken to restore outlook without AI"
"..."
"look at this 3-d game demo that claude one-shotted for me!"
Gareth Dwyer ⠕ retweeted
Been cooking!
Just left @rimelabs. Grateful to @lilyjclifford & team. Best models!
Now I built @alloglotapp, a language app with audiobooks from basics to Nietzsche, Don Quijote & Thoreau.
I built it with an automated software factory. I just type broken English. More soon!
Gareth Dwyer ⠕ retweeted
Never read an Airbnb listing again. Tell Jev what you want in plain English, and it reads 1,000 Cape Town listings and hands you the 10 best, scored out of 10, in about a second. Right inside Google Sheets 🏠
(trying something new with our demos!)
Gareth Dwyer ⠕ retweeted
I asked Fable 5.2 to "go wild" making a game, and all I got was this Pong clone.
And then I played it ...
👇
Gareth Dwyer ⠕ retweeted
Having built many dozens of Jev-powered projects and demos over the past few months, I can say with confidence my #1 advice for how to use Jev is:
Just put Jev everywhere.
Every time I think of a new (not Jev specific) project, I think to myself "what could Jev do in this system to make it cooler/smarter/weirder/sillier/whateverer" and then I do it.
The objective is not to use it in places where it makes total sense. The objective is to continue to expand my mind to see Jev the way our long term vision has always been: as an enabler of a new kind of smart software. The more you train your brain to truly think of Jev as smart primitives that can slot in to code in anywhere a primitive could live, the more you find novel use-cases.
Don't stop building. Don't stop playing. Experiment forever and ever and have fun with it!
Gareth Dwyer ⠕ retweeted
I'm starting to sense a divide in the tech industry:
- People thriving on autonomy who are reinventing how they work and what tools they use with AI: LOTS of work, but they're enjoying it and getting very efficient
- People used to being mentored / trained / or simply told what to do and how. Frustrated with the lack of guidance from management on what to do with AI; also frustrated by seeing some of their peers thriving; increasingly bitter about AI and the state of the industry, as a whole
Neither group seems to be aware that there's another group, or have empathy for the other group
Gareth Dwyer ⠕ retweeted
In a Discord town hall a few weeks back, I talked about Intelliprompter: my first-ever AI-integrated personal project.
I had shelved it due to LLM costs, but revived it the day I first learned about Jev.
We now have a Cookbook up on our docs showing how to make your own! 🔗 ➡ 🧵
Gareth Dwyer ⠕ retweeted
The South African digital risk matrix is flashing red today.
Here is what is breaking in our tech ecosystem right now:
/ @FNBSA suffered a massive outage this morning, locking users out of their mobile app, online banking, and card payments. As we aggressively push toward a cashless society, this is a stark reminder that our digital financial infrastructure requires absolute and zero-downtime resilience.
/ The Labour Court @OCJ_RSA has just drawn a line in the sand regarding "Shadow IT" and POPIA. A highly paid employee was fired, and the dismissal upheld, simply for forwarding work emails to her personal @gmail account. The court ruled that organisations handling bulk personal data are completely entitled to enforce zero-tolerance security policies to prevent data leaks. Your casual workarounds are now a dismissible cybersecurity threat. Tjo!
/ Speaking at the Singularity SA Summit @SUSouthAfrica, Google Cloud's @googlecloud VP delivered a harsh truth to local executives that we're entering the third wave of agentic AI, but deploying it merely to accelerate outdated, broken bureaucratic processes is a disaster. We must stop buying AI as a band-aid and start redesigning our workflows from the ground up.
This Google message resonates with my MBA teachings on digital transformation because true DX is not only about adopting frontier tech. It is largely about re-engineering your workflows and operating model first (the transformation), before deploying digital technologies such as AI.
AI is an amplifier of good or bad workflows, so think before you deploy.
Gareth Dwyer ⠕ retweeted
I never iterated this fast on any software ever, read how @pidotdev and pi-durable enable this:
danielgriesser.com/posts/ult…
Gareth Dwyer ⠕ retweeted
I really hope this changes course back on track.
Gareth Dwyer ⠕ retweeted
Interesting survey, fill it out if you have a couple mins
Have you ever let AI write (or rewrite) any part of your tech blog posts? If so, please answer these quick questions, totally anonymously. It should take less than 3 minutes.
forms.gle/3zYuANeuCdcmFt9D9
i love writing. i never let agents write for me. whether it's docs, or tweets, or shitposts, or even prompts (okay maybe that one's harder as agents manage other agents)
there's a simple reason for that: i still strongly believe that writing is thinking. it's a skill i hope to never give up. it's also important to me that whatever i say is always authentic, mistakes and all. my favorite posts are the ones that i just full send without any editing
i suspect it also helps with prompting. being able to clearly express what i want, and knowing the vocabulary i need to be able to direct my agents does seem to make for more effective usage of agents. in the same way that engineers have an advantage when we prompt agents to write code, because we are intimately familiar with the words that make up the discipline
knowing the right word incantations to get an agent to produce a desired goal almost feels like spellcasting. as models get more capable, and that input box looks more and more like a magic box that can output anything, words become the interface between you and the genie
there may come a time when we give up the keyboard in exchange for brain-computer interfaces. but until then, i'm writing every word.





























