Are the funding claims checked, or repeated from the launch post?

The entry copy says the video covers funding-skepticism reporting in The Register rather than only vendor material, which is the more interesting half. It does not document the specific figures, so the primary claims live in the cited reporting, not here.

Does it reproduce the hands-on test's numbers, or just cite it?

Cites it. The description points at an Every.to hands-on test alongside the launch posts, and the entry copy says this video summarizes other outlets and that you should go to those pieces for the primary measurements. The source page carries no transcript or figures to verify.

What does it say about answers outside the supplied options?

That is the catch the entry copy records: the model cannot pick an answer that is not in the set you gave it. How often that happens, and how you would catch it from the confidence value, is not documented in the video or the entry copy.

Is the parallel speed argument measured against an LLM?

Not in anything recorded here. The entry copy covers typed decisions with confidence scores, parallel output, and the Doom demo, but documents no latency comparison against a generative model, and the source page is title-only. The Doom demo is illustrative.

Who made the video, and what sources does it cite?

Heroic steps published it, citing TypeSafe, The Register, and Every.to. The entry describes it as a summary of those outlets rather than primary reporting.

What demo does the clip include?

A Doom demo, listed alongside typed decisions with confidence scores and parallel output speed.

Where should I go for the primary measurements?

The cited outlets. The entry's caveat is that this video summarizes other pieces, so rewatch those for the original numbers rather than relying on the recap.

CommunityJev Media & Talks

This AI Can't Write a Single Word. That's the Point.

Heroic steps on typed decisions with confidence scores, parallel output speed, the Doom demo, and citations to TypeSafe, The Register, and Every.to.

See highlights

Open youtube.com

Format: Video
Published by: Community
Author: Heroic steps
Last verified: Sep 2026

Tagscommunity video explainer evaluation

Works forJev for Classification

Used inJev for Software & DevTools

Highlights

Description cites an independent Every.to hands-on test alongside launch posts.
Covers funding skepticism reporting in The Register, not only TypeSafe marketing.
Calls out the catch when the right answer is not in the supplied option set.

Watch out

Summarizes other outlets; rewatch those pieces for primary measurements.

Questions about this entry

Short answers drawn from the linked project or page.

Reactions & coverage

Posts, threads, and videos about this entry from around the web.

YouTube: Launch summary citing Every.to hands-on testing, Register skepticism, and the Doom demo.

Hacker News: Simon Willison's overview of Noul, Choice, and Score, and where a decision model fits.

benwerd53 pointsJev introduces a new shape of LLMRead the thread on Hacker News

Hacker News: swyx posts the Latent Space interview with Diogo Almeida.

swyx4 pointsJev: System One Models for Prod, Not God – With Diogo Almeida, CEO, TypeSafe AIRead the thread on Hacker News

X: Diogo Almeida in five minutes

Related terms

Glossary definitions related to this entry.

Precision and recall

More Jev resources like this

Jev by TypeSafe AI, a System-1 decision model

CampusX on Jev as a classifier over schema-constrained choices, with a live benchmark and an e-commerce review demo.