jev-orchestrator
Python library to batch, gate, trace and calibrate thousands of typed Jev decisions, maximising questions per call.
Overview
jev-orchestrator is a workflow orchestration library for System One models. The README argues that existing frameworks such as LangChain, DSPy, Temporal and Prefect assume model calls are slow, expensive and return strings, whereas a Jev call returns typed values quickly and evaluates every question in parallel and in isolation. So the goal flips: maximise questions per call, then focus on knowing which decisions can be acted on without a human.
The library implements what the README says TypeSafe documents only in prose: multi-stage workflows, fan-out across records, confidence-gated escalation, staying under token budgets, cost accounting and proving thresholds hold. Questions from independent branches are merged into fewer HTTP requests; a demo shows five decisions across three branches sent in one call. The core has zero dependencies, an optional extra adds the official SDK, and a deterministic offline backend lets you explore everything without a key. Status is early and unstable.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which category does this support message belong to?(inferred) | — |
| Yes / no | Is this a billing problem?(inferred) | yes · no |
| Score | How urgent is it?(inferred) | — |
How it works
- Define a run and ask typed questions via Choice, Noul and Score
- The engine batches questions across branches into few requests
- Answers are gated by confidence and escalated when needed
- Costs and traces are recorded for calibration checks
Strengths
- Automatic batching of many questions per request
- Zero-dependency core with offline backend
- Built-in cost accounting and tracing
Caveats
- Early and unstable; API will change before 1.0
- Calibration tools rely on you supplying labelled data
Getting started
pip install jev-orchestrator, then run jev demo (no key needed).At a glance
- Type
- Project
- Status
- prototype
- Language
- Python
- Runtime
- Library and CLI
- Has code
- yes
- Has tests
- yes
Use cases and integrations
How curated lists describe it
Workflow orchestration for System One models.
wh000wh000/awesome-jev-live
Frequently asked questions
Why a new orchestration library?
The README argues existing frameworks minimise calls, while System One models reward packing many questions per call.
Can I try it without a key?
Yes. A deterministic offline backend runs the demo and examples.
Is it stable?
No. The author marks it early and says the API will change before 1.0.
Related projects
Open-source workspace for building, deploying and monitoring AI agents and workflows, listed as integrating TypeSafe System One.
ProjectproductionTypeScriptWorkflow DSL for existing agents that uses Jev for the steps that are decisions and calls an agent only when investigation is needed.
LibrarybetaTypeScriptn8n community node for Jev that adds a typed smart if-statement: choice, score and noul with probabilities in workflows.
ProjectbetaTypeScriptn8n Jev Classification node that branches items by category with a Needs Review branch, plus parallel and batched requests.
ProjectbetaTypeScriptn8n node that sends a state and a map of typed questions to /v1/systemone in one request and returns per-question answers.
ProjectbetaTypeScriptn8n community node for TypeSafe's System One API with yes/no, choice and score questions, retry handling and an AI Agent tool.
ServicebetaTypeScript