# System One Models

> Independent hub for System One models: AI models that return typed, calibrated decisions instead of text. The first is Jev by TypeSafe AI. This site catalogues use cases, real examples, decision recipes, explainers, and a glossary. It is not affiliated with TypeSafe AI or any model vendor.

Contact: https://systemonemodels.org/submit/

## Models
- [Jev](https://systemonemodels.org/models/jev/): The first System One model: typed answers and calibrated probabilities, never generated text.

## Guides
- [Choice, Score and Noul: the three primitives](https://systemonemodels.org/guides/choice-score-noul/): Choice, Score and Noul are the three question types a System One model answers. Concrete examples, their limits, and how confidence is derived.
- [How to build with System One models](https://systemonemodels.org/guides/how-to-build-with-system-one-models/): State in, typed questions out. Code owns the thresholds and side effects. The four official patterns, and the jaggedness list as design constraints.
- [How to get Jev API access](https://systemonemodels.org/guides/how-to-get-jev-access/): How to get Jev API access: the TypeSafe waitlist, the console key page, the TYPESAFE_API_KEY env var, SDK installs, and a first working call.
- [Is Jev just a zero-shot classifier?](https://systemonemodels.org/guides/is-jev-just-a-classifier/): Engineers called Jev a zero-shot classifier within hours of launch. What the comparison gets right, what it misses, and what is actually new.
- [Jev architecture: what is known and what is not](https://systemonemodels.org/guides/jev-architecture/): What TypeSafe has said about Jev's architecture: parallel output in one pass, typed answers, RLCD. What it has not said: weights, parameters, design.
- [Jev explained: TypeSafe AI's decision model](https://systemonemodels.org/guides/jev-explained/): Jev is TypeSafe AI's System One model: it returns typed decisions and calibrated probabilities, not text. Specs, pricing, access and limits.
- [Jev speed and pricing](https://systemonemodels.org/guides/jev-speed-and-pricing/): Jev costs $0.042 per million input tokens with free output. TypeSafe claims 70ms to 500ms and 40x to 200x. What those numbers cover.
- [RLCD explained: training a model to be calibrated](https://systemonemodels.org/guides/rlcd-explained/): RLCD is TypeSafe's training method for Jev: Reinforcement Learning for Calibrated Decisions. How it differs from RLHF and RLVR, and what is public.
- [System One models vs LLMs](https://systemonemodels.org/guides/system-one-vs-llm/): System One models return typed decisions; LLMs return text. A side-by-side on output, latency, cost, hallucination and when each one wins.
- [What is a System One model?](https://systemonemodels.org/guides/what-is-a-system-one-model/): A System One model returns typed, calibrated decisions instead of text. What that means, where the name comes from, and how Jev fits.

## Use cases
- [Agent routing and skill selection with System One models](https://systemonemodels.org/use-cases/real-time-and-agents/agent-routing-and-skill-selection/): An agent with a large skill roster picks badly from truncated descriptions. Jev ranks the whole roster in one call and can also answer that nothing fits.
- [Citation verification with System One models](https://systemonemodels.org/use-cases/retrieval-and-knowledge/citation-verification/): Check whether a quoted source actually backs the claim built on it. A string match catches fabrications, then one Jev Choice reads the surrounding context.
- [Compliance verification with System One models](https://systemonemodels.org/use-cases/safety-and-quality/compliance-verification/): Check a policy or contract against a written rulebook. Jev answers a whole compliance checklist in one typed request and flags the uncertain findings.
- [Composite scoring with System One models](https://systemonemodels.org/use-cases/data-and-operations/composite-scoring/): Break one fuzzy judgment into separate rated dimensions. Jev scores each one, and your code holds the weights, so you can see how a ranking was built.
- [Confidence-gated actions with System One models](https://systemonemodels.org/use-cases/workflow-control/confidence-gated-actions/): Use Jev's confidence value as a second axis: the answer says what the user wants, confidence says whether your code should act on it without asking.
- [Entity alignment with System One models](https://systemonemodels.org/use-cases/retrieval-and-knowledge/entity-alignment/): Decide whether two records describe the same thing using one Jev Score whose levels are the outcomes: merge, leave unlinked, or send to a curator.
- [Feature extraction for machine learning with System One models](https://systemonemodels.org/use-cases/data-and-operations/feature-extraction-for-ml/): Turn free text into numeric columns a gradient-boosting model can train on. Each Jev question becomes a feature, and the probabilities carry the uncertainty.
- [Hierarchical classification with System One models](https://systemonemodels.org/use-cases/data-and-operations/hierarchical-classification/): Walk a deep taxonomy one level at a time. Each node is a Jev choice question, and the probabilities let you keep several paths open instead of one.
- [Intent and model routing with System One models](https://systemonemodels.org/use-cases/workflow-control/intent-and-model-routing/): Classify what a request is and how hard it looks with one Jev call, then send it to plain code, a small model, a frontier model or a person.
- [LLM guardrails with System One models](https://systemonemodels.org/use-cases/safety-and-quality/llm-guardrails/): Screen every message into and out of an LLM app with one Jev call: a probability per hazard, a severity rating, and routing thresholds you own in code.
- [RAG passage filtering with System One models](https://systemonemodels.org/use-cases/retrieval-and-knowledge/rag-passage-filtering/): Screen every retrieved passage with Jev before it reaches the answering model: keep the usable ones, flag contradictions, drop injected instructions.
- [Real-time control with System One models](https://systemonemodels.org/use-cases/real-time-and-agents/real-time-control/): Jev answers fast enough to sit inside a game loop or a UI frame. Your code serialises the current state, asks a typed question, and acts on the answer.
- [Self-consistency checks with System One models](https://systemonemodels.org/use-cases/safety-and-quality/self-consistency-checks/): Ask Jev the same question repeatedly and see how far the answer moves. Measuring that spread tells you where a threshold is safe and where it is not.
- [Semantic code linting with System One models](https://systemonemodels.org/use-cases/safety-and-quality/semantic-linting/): Run your team's written conventions as Jev questions in CI. A System One model reads each diff hunk and returns typed judgments your pipeline can act on.
- [Semantic reranking with System One models](https://systemonemodels.org/use-cases/retrieval-and-knowledge/semantic-reranking/): Score each query and candidate pair with one Jev question, then sort the shortlist by that number. TypeSafe's legal retrieval test moved top-1 from 5% to 18%.
- [Structured extraction with System One models](https://systemonemodels.org/use-cases/data-and-operations/structured-extraction/): Jev does not write text, so extraction works in reverse: code finds candidate values, Jev picks the right one, and the value you get back is a verbatim copy.
- [Support inbox triage with System One models](https://systemonemodels.org/use-cases/workflow-control/support-inbox-triage/): Triage support tickets in one Jev call: category, severity, refund intent and frustration come back as typed values your routing code can branch on.
- [Typed tool dispatch with System One models](https://systemonemodels.org/use-cases/workflow-control/typed-tool-dispatch/): Pick the tool and fill its arguments with Jev Choice questions over the values each function already accepts, then let your own code make the call.

## Recipes
- [Ask four questions in one call and route in code](https://systemonemodels.org/recipes/parallel-questions-fan-out/): Send a choice, two nouls and a score to Jev in a single request, including questions that only matter for some outcomes, then pick a tool path in code.
- [Check a citation against its source with a noul](https://systemonemodels.org/recipes/citation-check-noul/): Catch fabricated and unsupported citations with Jev: a string match finds missing quotes, then one noul returns the probability the section backs the claim.
- [Gate a destructive action on confidence](https://systemonemodels.org/recipes/confidence-gated-action/): Ask Jev what the user wants and whether they confirmed it, then let per-action confidence thresholds decide what runs, what asks again, and what waits.
- [Re-rank retrieved passages with a Score rubric](https://systemonemodels.org/recipes/rerank-passages-score/): Score each query-passage pair against a five-level relevance rubric, sort the shortlist by the number Jev returns, and cut the weak tail in your own code.
- [Route a support ticket with one Choice](https://systemonemodels.org/recipes/support-triage-choice/): Send a support ticket to Jev as state, read back one queue label and a confidence value, and let your own router decide which tickets get assigned.
- [Score a sales lead on four dimensions and weight them yourself](https://systemonemodels.org/recipes/composite-lead-scoring/): Ask Jev four separate score questions about a lead, normalise each to 0 to 1, then combine them with weights your code owns instead of one vague quality rating.
- [Screen a message for prompt injection with Nouls](https://systemonemodels.org/recipes/prompt-injection-screen-noul/): Run four Noul questions over one message in a single Jev call, get a probability per hazard, and threshold them in your code to pass, strip, review or block.
- [Walk a taxonomy with one choice per level](https://systemonemodels.org/recipes/hierarchical-classification-cascade/): Classify into a deep taxonomy by asking Jev one choice question per level, multiplying the edge probabilities and stopping the walk when confidence drops.

## Glossary
- [Calibrated Probability](https://systemonemodels.org/glossary/calibrated-probability/)
- [Calibration](https://systemonemodels.org/glossary/calibration/)
- [Choice](https://systemonemodels.org/glossary/choice/)
- [Composite scoring](https://systemonemodels.org/glossary/composite-scoring/)
- [Confidence](https://systemonemodels.org/glossary/confidence/)
- [Confidence Gating](https://systemonemodels.org/glossary/confidence-gating/)
- [Decision Model](https://systemonemodels.org/glossary/decision-model/)
- [Expected calibration error](https://systemonemodels.org/glossary/expected-calibration-error/)
- [Hallucination](https://systemonemodels.org/glossary/hallucination/)
- [Intent routing](https://systemonemodels.org/glossary/intent-routing/)
- [Jaggedness](https://systemonemodels.org/glossary/jaggedness/)
- [Jev](https://systemonemodels.org/glossary/jev/)
- [Non-autoregressive](https://systemonemodels.org/glossary/non-autoregressive/)
- [Noul](https://systemonemodels.org/glossary/noul/)
- [Question](https://systemonemodels.org/glossary/question/)
- [RLCD](https://systemonemodels.org/glossary/rlcd/)
- [Score](https://systemonemodels.org/glossary/score/)
- [Speculative fan-out](https://systemonemodels.org/glossary/speculative-fan-out/)
- [State](https://systemonemodels.org/glossary/state/)
- [Structured outputs](https://systemonemodels.org/glossary/structured-outputs/)
- [System One Model](https://systemonemodels.org/glossary/system-one-model/)
- [System Two Model](https://systemonemodels.org/glossary/system-two-model/)
- [Thinking, Fast and Slow](https://systemonemodels.org/glossary/thinking-fast-and-slow/)
- [Typed output](https://systemonemodels.org/glossary/typed-output/)
- [TypeSafe AI](https://systemonemodels.org/glossary/typesafe-ai/)
- [Zero-shot classifier](https://systemonemodels.org/glossary/zero-shot-classifier/)

## Examples directory
- [All examples](https://systemonemodels.org/examples/)
- [Projects](https://systemonemodels.org/examples/projects/): 87 entries
- [Tools](https://systemonemodels.org/examples/tools/): 110 entries
- [Cookbooks](https://systemonemodels.org/examples/cookbooks/): 25 entries
- [Videos](https://systemonemodels.org/examples/videos/): 14 entries
- [Articles](https://systemonemodels.org/examples/articles/): 35 entries
- [Discussions](https://systemonemodels.org/examples/discussions/): 23 entries
- [Lists](https://systemonemodels.org/examples/lists/): 7 entries
- [Alternatives](https://systemonemodels.org/examples/alternatives/): 22 entries

## Optional
- [About](https://systemonemodels.org/about/)
- [Submit an example](https://systemonemodels.org/submit/)
- [HTML sitemap](https://systemonemodels.org/sitemap/)
- [Sitemap](https://systemonemodels.org/sitemap-index.xml)