Eugene Shvarts

building the future of ai @typesafeai

[Eugene Shvarts @mathfax]

fun fact: we almost put our office in Oakland

[Nathan LeClaire @dotpem]

little known fact jev is from oakland actually

[Eugene Shvarts @mathfax]

@dotpem jevbot ilu

[Eugene Shvarts @mathfax]

@ainativefirm pick your favorite heuristic! You can use individual nouls over each entry if you don't trust the partition variance in a tournament

https://t.co/74tF0GEq9R

[Eugene Shvarts @mathfax]

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!)

[Eugene Shvarts @mathfax]

One of the dreams (pun intended) of smart software is that if you can describe it, you can build it!

Why should detecting people at the edge of the video be any harder than just asking?

[Noman @Nomandsign]

Introducing Wisp, open-source content-aware video player that softly fades the edges of a video into the background for a dreamlike look.

It uses Jev by @typesafeai to detect people, animals and faces and keeps them sharp as they get close to the edges.

[Eugene Shvarts @mathfax]

Steal the sheet, drop in your key, and throw whatever you want at it:

https://t.co/ZrbvXdRUzv

[Eugene Shvarts @mathfax]

1,000 requests → down to 11

30+ seconds → about 1 🔥

Half a cent per search.

[Dan Robinson @danlovesproofs]

Nice unlock from jev: we can show you what parts of your codebase are relevant to a vuln scan and let you fine-tune the targeting.

The classifications are fast enough and cheap enough that you can support "insert common sense here" functionality inside of a UX that needs to be snappy. This works for repos with 100k+ files.

[Eugene Shvarts @mathfax]

Highlights from @CompleteSkeptic ‘s latest talk!

[Turing Post (Ksenia Se) @TheTuringPost]

“Software is gonna be so fucking back.”

.@CompleteSkeptic is full of bangers and you will have to listen to his full talk at @aiDotEngineer, but for now, here is our top-10 of his pearls:
1“We’ve been announcing AGI three months away for two years now.” Meanwhile, go outside San Francisco and a lot of the world looks unchanged. Where the fuck is all the automation?
2“You get Einstein 95% of the time, but Mr. Bean 5% of the time.” And if you can’t tell which is which, how are you supposed to automate anything?
3“I do believe RLHF is part of the problem.” His complaint: we optimized models for assistance, with a human checking the result, and then expected them to reliably do work on their own.
4“The bitterest lesson is doing the right task.” Diogo thinks that compute matters more than algorithms, data matters more than compute, and choosing the right task matters most. You get what you optimize for.
5“I think the world still has LLM Stockholm Syndrome.” We keep designing around the limitations of today’s models. Why is the best we can come up with still the same software, cheaper to write, with a chatbot on the side?
6“What if we can program with common sense, with intention, with human understanding?” Very passionate talk! He suggests: put intelligence into the software’s inner loop. What could we build if understanding intent became a building block we could actually rely on?
7“Real fucking engineering by people who care a lot about the end result.” Break tasks down, build stages you can verify, and actually manage state. Why should every agent rediscover the same codebase patterns over and over?
8“There’s an asymmetric advantage for people who are, like, bad actors and don’t give a shit.” That’s his explanation for slop: unreliable output is much easier to tolerate when everyone else pays the cost of reading, checking, and filtering it.
9“I call this software research. That’s not a joke.” Explore what software could become, including weird things with no obvious business case. Build something because you can. Maybe one percent of those ideas leads to something awesome.
There is going to be SaaSopalooza! (I don’t know how to write this word but you got me)
10“Software is gonna be so fucking back.” The future he wants is one where “can we automate this?” becomes “how much work will it take?”

[Kunal Abichandani @kunalabcdani]

What if we could manipulate UI at the speed of thought?

Built this voice coffee ordering system with @typesafeai Jev. The drink and any updates show up while I’m still mid-sentence, ~200 ms from word to screen.

Feels pretty magical!

[Scott Hanselman 🌮 @shanselman]

What happens when you combine embedding models with System 1 decision models? You get the intersection, which is VENN —> Vector Decision Models, a very cool OSS research project from Mark Russinovich that I was fortunate to review early and am excited to support.

[Allie the Icon @allietheicon]

Leading up to launch, I tried to come up with a demo vid that would appeal to Excel nerds, as we know Jev has a lot of potential in the BI/data analysis world. I wasn't able to come up with an idea that presented visually enough.

@phil_renaud has done it! This so SO COOL

[Diogo Almeida @CompleteSkeptic]

@swyx

[Tejas Manohar @tejasmanohar]

our eng team got #1 on HN today!

they tested if fast + cheap decision models like Jev @typesafeai could grade complex data-matching tasks as well as frontier models.

Jev with Claude Opus got close to Opus-level accuracy for a fraction of the cost.

super cool. post below:

[Nick Shirobokov @nicksherrow]

Can't stop thinking how much stuff Jev is doing to save AI engineers asses from stupid systems design patterns while the rest of the domain keeps pushing tool calling loops everywhere

[Eugene Shvarts @mathfax]

@sheherenow_ want want want

[Allie the Icon @allietheicon]

Why don't we at TypeSafe call Jev a decision model?

If you think of Jev's questions as making *decisions*, I believe you'll use it at the wrong layer of abstraction and your reliability will suffer.

I wrote all about this on our site 🧵👇

[Jacob Effron @jacobeffron]

Why hasn't AI automated more work? What's holding coding agents back? And are the labs' valuations built on the right expectations?

@CompleteSkeptic, co-founder and CEO of @typesafeai, was the perfect person to discuss this and more. Diogo coauthored OpenAI's InstructGPT paper, and a few weeks before we recorded, TypeSafe launched Jev, a "system one" model that returns structured decisions instead of text.

It was a ton of fun to discuss this with Diogo.

The conversation includes:

0:00 Intro
1:14 The Early Bet on RLCD
4:40 Built for Humans vs Computers
6:39 Roadmap & Model Limitations
11:04 Mental Models of Models
16:07 Do LLMs Have a Place in 10 Years?
17:32 New Model Shapes
21:24 Problems Still Unsolved
27:00 The Tyranny of the KV Cache
31:21 Rethinking Coding Agents
34:40 Multimodality
37:52 Images, Games & Computer Use
39:36 Could This Have Been Built 5 Years Ago?
42:01 Benchmarks & Copycat Releases
44:12 Fundraising
45:27 Quickfire

Listen here:

YouTube: https://t.co/K8kp8aFv3T
Spotify: https://t.co/GImUYej9It
Apple: https://t.co/KcuUJNkT3Q