We are a community site for Jev AI extras — widgets, workflows, and ideas you can try in the browser and share with others. Official docs and API live with TypeSafe AI; we focus on what you can build and play with today.
Pick something to try, run it online when we host it, then fork the idea or submit your own write-up. No install required for most demos.
Explore playbooks, mini-tools, and feature notes from other builders — sorted by what you can try online first.
Launch hosted demos in one click when available. Tweak inputs, see structured Jev-style outputs, and learn by doing.
Publish your variant, write up a trick, or request a hosted demo. The catalog stays useful because people contribute.
import typesafe client = typesafe.Client(api_key="jev_...") decision = client.decide( model="jev-1", state=support_ticket_text, # example snippet from a community playbook questions={ "intent": ["refund", "bug", "billing", "other"], "needs_human": bool, "urgency": ("score", 1, 5), }, ) # → {"intent": ("refund", p=0.94), "needs_human": (false, p=0.88), # "urgency": (4, confidence=0.91)} · ~120ms end to end
TypeSafe AI ships the model, core infrastructure, and raw API. JevAI Community collects peripheral features, curated API endpoint recommendations, mode-driven recipes, and community play patterns — complete with live online testing and ready-to-use API implementations where we can.
| Official Jev / TypeSafe | JevAI Community | |
|---|---|---|
| Focus | Model, infrastructure, and native API execution. | Peripheral tools, API endpoint recommendations, and shared playbooks. |
| Content | Docs, pricing, SDK references, and enterprise access. | Community write-ups, API recipes by mode (compaction, router, etc.), and live demos. |
| Experience | Integrate via production-ready API in your stack. | Browse, copy API snippets, and test different modes in the browser. |
| Contributions | Product team roadmap and core engine updates. | User-submitted API use cases, community plugins, and featured workflows. |
Content is contributed and curated by the community. We organize API endpoint recommendations, developer workflows, and tools into the following categories:
Small, interactive UIs to test Jev-powered flows right in your browser without writing code—try out classifiers, scorers, and routers with millisecond latency.
Step-by-step patterns: define input states with typed schemas, leverage different mode parameters, and learn how to wire structured results into your apps or workflows.
Community-driven glue and production-ready plugins—such as fast-jev-compaction for Claude Code, custom webhooks, dashboards, and developer helpers that sit alongside the official API.
Creative community demos (ranging from real-time workflows to playful UX) featuring deep-dive notes on latency optimization, prompt design, and structured JSON outputs.
A community-driven repository of battle-tested prompts and typed schemas (TypeScript / JSON Schema). Copy-paste ready templates optimized for classification, extraction, and routing to ensure robust zero-text reliability.
Discover what's next for the community and open-source ecosystem. Learn how to submit your own playbooks, contribute plugins like fast-jev-compaction, or participate in community-driven feature requests.
From coding agents to real-time AI and visual inference, developers are finding new ways to use the Jev Model for fast, structured decision-making.
Replace lossy summaries with Jev-powered decisions.
A Claude Code plugin that uses Jev to decide which tool calls and results are still relevant. Stale context can be dropped or truncated, while everything kept remains verbatim — helping preserve important details during long coding sessions.
tamaratran/fast-jev-compaction
Community Claude Code plugin — fork, inspect, and wire your own Jev API key.
View on GitHubTurn unstructured input into structured decisions.
Use Jev to identify user intent and route requests based on the signals that matter. Instead of generating a long response, Jev can directly answer the decision your application needs.
Make real-time AI interactions more actionable.
Combine Jev with GPT Live to make fast decisions during live conversations — including identifying relevant products and recommending what to show or suggest next.
Review and classify invoices in seconds.
Use Jev to process large numbers of invoices and make structured accounting decisions, helping automate tasks such as categorization and financial workflow routing.
Bring Jev-style decision making to visual tasks.
An educational visual inference experiment on Apple Silicon that explores shared context, direct candidate scoring, and local visual demos — including three-color product sorting.
vinnylarouge/jevlike
Visual scoring demos and a Jev-like starter you can run locally on Apple Silicon.
View on GitHubTell us what you built around Jev — demos, integrations, or a clever workflow. We'll feature the best online.
Submit your idea