Finding signal on Twitter is more difficult than it used to be. We curate the best tweets on topics like AI, startups, and product development every weekday so you can focus on what matters.
Named in these same posts. This does not imply a comparison or recommendation.
How this is put together
Public posts from the accounts Tech Twitter monitors, in the selected window. Findings need three supporting authors and a published source. Announcements can cite one known-affiliated account. This is a sample of the conversation, not a survey or a measure of adoption.
The posts behind the picture
Public source posts
@exm7777
Jev can't write a single word, which is exactly why it should pick what my agents read
my Obsidian vaults hold 5k+ notes, one idea per note, each with a one-line summary in an index
but Claude can't read all of that on every turn, and a keyword search misses the note that makes the same point in different words
so Jev does the picking:
> step 1: Jev reads the 28 vault descriptions and picks where the task belongs
> step 2: it checks every note summary in that vault in one call, each judged alone
> step 3: the top 5 come back with full text - Jev drops any that miss the task
> step 4: my code loads the survivors into Claude's context (and nothing else)
every pick comes from a list i define, so Jev can't invent a note
the worst it can do is pick the wrong one... and each one carries a confidence score that tells my code when to fall back to Claude
I've been using Jev for all kinds of things. This morning I had a realization I kind of like:
Use it to make non-black-box embeddings.
Instead of an embedding model spitting out 1,536 numbers that mean nothing, you ask Jev questions about each document. The answers become the vector.
An email in Cora:
"I got charged twice this month, pls fix asap"
[is_customer, urgent, about_billing, needs_reply]
[1.0, 0.9, 1.0, 1.0]
A newsletter:
[0.0, 0.0, 0.0, 0.1]
A friend asking about lunch:
[0.0, 0.1, 0.0, 0.7]
Then it's just old-school cosine similarity search. Search "billing issues from customers" as [1, 0.5, 1, 0.5] and the double charge comes out on top.
Same idea for our articles at Every:
[is_tutorial, about_ai, contrarian, beginner_friendly]
Or support tickets:
[is_bug, angry, churn_risk, enterprise]
Every number has a name, so you can see why something matched. Need a new dimension? Add a question. Want urgent stuff first? Change the query vector.
Trying this in @CoraComputer now to make search fast.
Deel put Jev head-to-head with frontier LLMs across multiple use cases: up to 59× cheaper, up to 4× faster, and quality at parity or better on 6 of 8 checks.
TypeSafe opened Jev to us on Sunday. Expected cheaper and faster, got both.
The real shock? Accuracy surged.
Repeat questions: 70% → 97%
Expense categories: 50% → 86%
Escalations: same catches, fewer false alarms
On Sept 30 we’re working with @supabase to put on HYPERSHIP DAY on Product Hunt.
If you think you can build and ship fast, and you haven’t launched recently, submit your launch by midnight Sept 30. We’ll be providing some cool prizes including Jev credits and swag!
And if you’re not launching: Mark your calendars to try new products, make feature requests, and watch the products evolve over the course of the day. 🔥🔥
Sep 30 is HYPERSHIP DAY on @ProductHunt and we're cosponsoring with @typesafeai, makers of Jev.
Haven't launched in the last 6 months? Submit by midnight Sep 30, then build and ship new features live throughout the day.
Not launching? Show up anyway — try new products, make feature requests, and watch them ship your ideas in real time.
Prizes: Supabase credits, swag, and a few secrets. Are you fast enough?
Sep 30 is **HYPERSHIP** day on Product Hunt.
If you think you can build and ship fast, and you haven't launched recently, submit your launch by midnight Sep 30.
Be prepared to build and ship multiple features in real-time. Appropriately, given the speed theme, we’ve partnered with @typesafeai, makers of Jev, a frontier model at real-time speed, and @supabase, the backend platform you know and love!
Prizes will include Jev and Supabase credits, swag boxes, and secret prizes we’ll announce later.
And if you're not launching: Mark your calendars to try new products, make feature requests, and watch the products evolve over the course of the day.
This is the day where we redefine what shipping velocity means.
Don't sleep on using Jev-as-a-Judge for agent evaluation.
This is one of the most impressive Jev use cases I have found so far.
Jev is a natural fit as a Judge, but it doesn't mean you use it everywhere.
Similarly, you shouldn't use frontier models for evals everywhere.
I'm running lots of tests on this atm, but early results point to an optimized flow (balancing accuracy and cost) that combines Jev and frontier models.
Concretely, use Jev in high-confidence situations, and escalate to a frontier model (GPT-6 or Opus 5.5) in low-confidence verdicts.
Entire write-up coming soon. Let me know if you have questions as I build the full guide.
- building a product with AI: 2 days
- finding 100 people who might actually pay for it: 2 months
and no, having an AI agent write 500 cold emails doesn't solve the problem.
you still need to find the RIGHT people.
that's what @treg_ai is going after
Jev + Treg lets your agent search across 60+ data providers and score potential leads:
apparently 200 leads can cost as little as $1.13 (🤯)
On the question of model labs copying Jev:
Speculatively speaking, I think decision models could be a category for a new type of model lab, or at least a good model for non-lab AI companies to offer under their brand. They don't have the same economics as frontier models, so the frontier labs may not be incentivized to put as much compute towards developing them as the neolabs would be (and this is certainly true if the data is very different - I'm not sure about that).
It should also be a very easy model for non-lab software companies to finetune/customize, without worrying as much about model debt. I expect that the labs will eventually get back into this down the road for small models, so it will depend on how quickly the neolabs can hire humans that face the customers to get data and launch customizations.
I do think that the frontier labs will improve the DX for their small models, but there will still be demand for out-of-family decision models to double check or steer the LLM's work (just like we use out-of-family models to review the code written by your main model)
best self hosted jev alternative rn? doesnt have to be stupid smart
not about $ it's about privacy i want each incoming beeper text message to be classified as urgent/not urgent without sending it somewhere
most of the Jev use cases i'm seeing on X are absolutely useless
no denying AI agents will lean on system one models hard in the upcoming months... but from my (short) experience it overcomplicates systems most of the time
i got it working in some SEO and content pipelines i run, and honestly that's about it... everywhere else a cheap flash model gets the job done
because it can't technically write text, you still need an LLM to process its outputs, and i hate stacking layers over layers for simple tasks
that being said, it's incredibly fast at computer & browser use, i think this is where it will shine most
excited to see where this goes, i genuinely think it ends up in every serious agent stack... but like everything you gotta let it breathe a little
We have seen such an immense swell of demand that we have to temporarily pause signups for Jev. We need to ensure quality of service for our existing signups, which will continue to function. We are working diligently to ensure open access to Jev for everyone as soon as we can. Thank you.
Jev, in its current form, is only good for non-critical decisions. The reason is that it's small and not reasoning. It’s also bad at handling non-English input. We could expect this to improve over time.
For example, in Todoist, it can suggest a project. That’s a great use case. A bad use case would be using it as a critical decision-maker in a workflow (e.g., auto-assigning the project).
Just sharing a few thoughts as we are working with these new types of models 👍😊
SITUATION EXPLAINED: Is Jev a new kind of AI, or just a very fast classifier?
• Diogo Almeida claims Jev is up to 200x faster and 400x cheaper than LLMs
• Jev doesn't write anything. It returns scores and probabilities, fast and cheap
• @tenobrus: "every flashy demo is something existing done 10x faster and cheaper and not actually functional, and the launch was incredibly misleading"
@theojaffee: "It's very, very early days. It's like the equivalent of GPT-2 to what we have now with Astra. And it's possible that there are some tasks for which something really cheap and really fast is better than frontier intelligence."
Recommended reading. This offers a solid set of ideas for where you can place Jev in your agent harness.
It's cool to see ideas I have shared before, like approval gates, MCP/tool-calling routing, model routing, dynamic subagent patterns, structured skills, and more.
Feed the doc to your agent and start exploring.
The Jev moment happening now is similar in energy to the OpenClaw moment that happened in January, and also to the Opus 4.5 moment before that. Developers are scrambling to find use cases for a new hot thing, and it's a sudden blooming of creativity.
Models have typically NOT optimized for specific use cases and have gone the other direction, generalizing over all of them. This could be a very important moment for the whole AI ecosystem if this turns out to be the beginning of other model skews that make the market much more diverse, such as compaction, summarization, extraction and more, all of which can now be used to optimize harnesses and decouple them from provider lock-in.
We'll see what happens over the course of the next few weeks when the model labs optimize their small models more or try new ways of branding and packaging them.
What I like about Jev: for years, large generalist models sucked almost all the oxygen out of AI.
But there are huge opportunities in much more specialized and customized models, built for specific tasks and languages and as a result, orders of magnitude cheaper, faster and more optimized. There are 3 million of them publicly available on @huggingface.
Let's build a much more diverse AI ecosystem!
2 hot takes here:
1) Jev represents a new family of models focused on super-fast and cheap decision-making. Prior to Jev, I don’t think most people knew this was possible. I.e., this seems like a pretty big deal, and it has a ton of use cases.
2) Marketing is as critical as building. Making something great and making people aware of it are equally important.
Especially 2) is hard for many devs to accept.
