Lesson 02 · Jev 101

How Jev works: state + typed questions

Forget prompt engineering for a moment. Every Jev call has the same shape: state in, questions in, typed answers out.

State (your context)

State is whatever the decision is about: a ticket, email, agent trace, document, game state, log blob—often unstructured text, sometimes structured program state (company emphasis). Official docs: a string, a JSON object, or an array of text. Text only — no image, audio, or video yet.

ChatGPT parallel: the stuff you paste into the prompt.
Difference: you’re not also asking Jev to write the answer in English.

Official jaggedness note: accuracy falls when you stuff large irrelevant state (“context rot”). Filter. Send only what the decision needs.

Three question types

TypePlain EnglishChatGPT prompt you might retire
ChoicePick from labels you define. Returns the option, a distribution, and confidence.“Classify as A/B/C. Return JSON.”
ScorePlace on an ordered scale / rubric you define. Returns a weighted score plus confidence.“Rate 1–5 severity.”
NoulYes/no-style probability (0–1). No separate confidence field — the probability is the signal.“Is this spam? Answer yes or no.”

Multiple questions can be evaluated together in one request (company: independently, in parallel). Official theme: speculative fan-out — ask more than you might need, let code select. Launch blog: Choice cardinality up to 255. Score docs: at least two levels; this 101 does not invent a ceiling.

“System One is TypeSafe's model for building AI-powered software, not agents.”

TypeSafe docs · how to build · fetched Sep 20, 2026 · Source

What you do not get back

Jev does not return essays, code, tool calls, or chain-of-thought prose. Official docs: “jev-1.13 is not trained to generate text.” If you need an explanation for a human, call an LLM after (or beside) the decision—or show your own UI copy.

“jev-1.13 is not trained to generate text.”

TypeSafe docs · jaggedness · reviewed Sep 17, 2026 · Source

“Jev is not a calculator. We strongly recommend implementing any mathematical logic in code.”

TypeSafe docs · jaggedness · Sep 17, 2026 · Source

Why this feels more like code

Your application defines the answer space before inference. That constraint is the product: TypeSafe claims the model never makes type errors; schema matching is guaranteed by constructionCompany claim. You still design for wrong decisions via confidence and escalation.

“No type errors: This would be an easy thing to falsify with just a single counter-example, but it is mathematically impossible.”

TypeSafe launch blog · Sep 15, 2026 · Source

“We want AI to work alongside existing software as a primitive that any programmer can invoke for semantic judgement and decisions, while still using code for what it’s best at: exact computation.”

TypeSafe manifesto · fetched Sep 20, 2026 · Source

Constraints worth knowing

  • Context cited in models.md: 64k for state + all questions; 32k for state + longest question
  • English is primary; CJK and other languages are accepted with lower accuracy
  • No customer fine-tune / LoRA; same weights for all accounts
  • Not trained on customer requests or responses (official models.md)
  • State is data, not a hostile user by default — add your own injection defenses

Optional code · official surface

Endpoint from the official quickstart. Do not freeze a request body here — copy the current schema fromdocs.typesafe.ai.

POST https://api.typesafe.ai/v1/systemone
Authorization: Bearer $TYPESAFE_API_KEY

# Python: pip install typesafe-sdk
# JS:     @typesafe-ai/sdk
# Pin:    jev-1.13.0  (aliases jev-latest / jev-preview can move)