Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

ha-conversation-jev

Home Assistant conversation agent where Jev classifies an utterance and either calls a light, climate or cover service or hands off to Grok.

Projectprototype
View on GitHub ↗

Overview

Jev Assist is a custom Home Assistant conversation agent. Jev classifies each utterance and either takes a fast path, calling a light, climate or cover service directly, or hands the request to the SpaceXAI Grok conversation agent. Questions in German and English live in a criteria file and are sent in one system_one call with explicit retry and timeout settings.

Setup through HACS needs a TypeSafe or Vercel-issued Jev key, stored on the config entry, and Grok access via device-code OAuth sign-in or an optional API key fallback; you then choose Jev Assist as the conversation agent in an Assist pipeline. The fast path avoids sending simple home-control commands to a large model. Manual installation by copying the custom component folder is also supported.

What the decision model decides

TypeQuestionAnswers
ChoiceIs this utterance a light, climate or cover command, or something for Grok?(inferred)light · climate · cover · grok

How it works

  1. User speaks or types to Assist
  2. Jev classifies the utterance in one call
  3. Simple commands call Home Assistant services directly
  4. Other requests go to the Grok conversation agent

Strengths

  • Fast, cheap path for common commands
  • German and English support
  • Single Jev call per utterance

Caveats

  • Requires both Jev and Grok access
  • Home Assistant only

Getting started

Add https://github.com/luxus/ha-conversation-jev as a custom HACS integration, install Jev Assist, enter keys and select it in an Assist pipeline.

At a glance

Type
Project
Status
prototype
Language
Python
Runtime
Home Assistant integration
Has code
yes

Use cases and integrations

How curated lists describe it

  • A Home Assistant conversation integration routing simple lighting commands to services and other requests to Grok.

    logicrw/awesome-jev-projects

Frequently asked questions

What does Jev decide?

Whether an utterance is a simple device command or should go to Grok.

Which devices take the fast path?

Lights, climate and covers.

Which languages?

German and English.

Related projects

  • jevXagent

    Transparent proxy that asks Jev to classify each coding-agent prompt and injects the decision into the agent's context before it generates.

    ProxybetaPython
  • openclaw-jev-plugin

    OpenClaw plugin where Jev decides whether a group-chat bot should reply and, optionally, which model tier should answer.

    PluginbetaTypeScript
  • typesafe-assist

    Home Assistant conversation agent that maps Assist commands to built-in intents using a Jev fan-out of typed questions.

    ProjectbetaPython
  • TypeSafe Conversation for Home Assistant

    Home Assistant conversation agent that resolves commands from one System One call with a fan-out of typed questions, asking when unsure.

    ProjectprototypePython
  • AgentScope

    Alibaba's AgentScope 2.0 agent framework, which added a JevClassifierModel for binary, choice and score decisions via the TypeSafe SDK.

    ProjectproductionPython
  • HiRoute

    Local-first routing engine for long-running agent tasks that uses a Jev decider to pick a model group per turn.

    PluginproductionRust, TypeScript