TaraHome AI Assistant - Setup Wizard

Choose AI Provider

Select which AI service you want to use

Claude (Anthropic)

Advanced reasoning, great for complex automations

GPT (OpenAI)

Versatile and widely used, good all-around choice

Ollama (Local)

Run AI locally, no API key needed

⚠️ Use a model with tool calling support (e.g. llama3.1, mistral-nemo)

Gemini (Google)

Google's latest AI models, great performance

OpenAI Compatible

OpenWebUI, llama.cpp, LM Studio, vLLM, LocalAI

API Credentials

Enter your API key

API Key

Ollama Host URL

Base URL

OpenWebUI: https://your-domain/api/v1 | llama.cpp: http://localhost:8080/v1

API Key (optional)

Configure Limits

Set usage limits and safety thresholds

Response Length

Controls the maximum length of AI responses. Higher values allow more detailed answers but increase API costs and response time.

Max Tokens per Response 4096

500-1500: Concise | 2000-4000: Balanced | 8000+: Detailed

Rate Limiting

Limits requests per minute to prevent runaway API costs. Applies across all users of this instance.

Requests per Minute 20

5-10: Conservative | 15-30: Normal | 40+: High throughput

Safety Guardrails

The AI evaluates each Home Assistant command with a risk score (0-100). Commands above this threshold are blocked, protecting against harmful automations like unlocking doors or disabling alarms.

Risk Threshold 70

Commands scoring above this value will be blocked

Balanced protection for most homes.

Connect Home Assistant

Enter your Home Assistant details

Home Assistant URL

Use your Home Assistant's IP address (e.g., http://192.168.1.100:8123) rather than homeassistant.local, which may not resolve inside Docker.

Long-Lived Access Token

Create one in HA: Profile → Long-Lived Access Tokens

Review & Save

Confirm your configuration

AI Provider

-

Model

-

Limits

-

Home Assistant

-

Access Password Confirm Password