What Is Ask Jev? The Chrome Extension That Puts Verdicts on Every Page

Ask Jev is a community Chrome extension that does the smallest possible thing with Jev, which is exactly why it's worth a page: highlight text on any webpage, pose a question about it, and get back the model's native output, a probability for each candidate answer, defaulting to the Choice API's YES / NO / UNCLEAR when you don't customize. It is the model's interface with zero ceremony attached, and as a result it's the fastest way to feel what a decision model is before you build anything.

The honest use cases are modest and real: sanity-checking a claim's shape while reading ("does this paragraph actually state a verifiable number?"), triaging your own drafts with a quick verdict, prototyping question wording against live examples before committing it to a pipeline, and teaching colleagues what "probabilities per choice" means in ten seconds of show-and-tell. That last one matters more than it sounds: half the confusion documented on the limitations page evaporates when someone watches a closed question return calibrated numbers instead of prose.

The honest cautions are the standing ones, plus one specific to extensions: whatever you highlight is what gets sent, so the data-minimization instincts apply to your own browsing, and a browser extension's permissions deserve the same squint as any other's. It's a community tool, not official; verify the publisher, per normal extension hygiene, and remember one-off verdicts in a popup are demos, while calibrated, versioned, gated verdicts are systems. The extension is the appetizer; the getting-started guide is the meal.

Frequently asked questions

Is Ask Jev made by TypeSafe?

No: community-built, per ecosystem listings. Official surfaces live at docs.typesafe.ai; the extension is the ecosystem being quick.

What does it cost to use?

It rides on API access, so per-verdict economics apply: casual use rounds to zero. Bring your own key per the extension's setup.

Can I use it for real work?

For prototyping questions and spot checks, absolutely; for anything repeated or consequential, graduate to a pipeline with gates and logs. Popups don't have audit trails.