Back to GitHub projects

Jev Lab: Jev guardrail and pre-evaluation demo for business agents

A Next.js demo and calibration study that uses TypeSafe AI's Jev as a real-time guardrail and LLM-judge pre-evaluator for Portuguese-language customer-support agents via Vercel AI Gateway and LangChain.

仓库明确以 TypeSafe AI 的 Jev 为核心,通过 Vercel AI Gateway 集成 Jev 的 /v1/evaluate 端点,提供可运行的 Next.js 护栏与预评估演示、LangChain 业务智能体,并包含 Python 校准研究,属于真实使用和深入研究 Jev 的项目,而非同名日本脑炎或空仓库。

· 0 starsOpen on GitHub