IONOS Agent Docs v1.0 just launched!⭐ Star us on GitHub
docsFrontend GuideStreamlitOverview

Streamlit Frontend

The Streamlit app provides a quick UI to:

  • Select an IONOS AI model from the dropdown
  • Chat with the intelligent ReAct agent
  • View real-time responses with dynamic web search and reasoning
  • Experience context-aware conversations with chat history

Run

a) Windows (PowerShell)

# In a new shell
cd frontends/streamlit-starter
python -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -r requirements.txt  # or: pip install streamlit requests
streamlit run app.py

b) macOS / Linux (bash/zsh)

# In a new shell
cd frontends/streamlit-starter
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt  # or: pip install streamlit requests
streamlit run app.py

Note: The frontend expects the backend on http://backend-service:8000 by default (Kubernetes). For local development, create a .env file with BACKEND_URL=http://localhost:8000.

Configure

  • Backend URL defaults to http://backend-service:8000 for Kubernetes deployment
  • For local development: Create .env file with BACKEND_URL=http://localhost:8000
  • IONOS_API_KEY is required in frontend .env for fetching inference models
  • Model selection supports both IONOS Hub inference models and Studio fine-tuned models
  • The agent automatically handles web search and reasoning without additional configuration

Features

  • Real-time Chat: Instant responses from IONOS AI models
  • Dynamic Web Search: Automatically searches the web when needed (inference models only)
  • Context Awareness: Remembers conversation history
  • Model Type Selection: Switch between Inference and Fine-tuned models
    • Inference Models: IONOS Hub models with web search capabilities (fetched directly from IONOS API)
    • Fine-tuned Models: Studio models for specialized tasks (fetched from backend /studio/models endpoint)
  • Model Selection: Choose from available models based on selected type

Adding New Fine-tuned Models

To add a new fine-tuned model:

  1. Add model UUID to backend .env:
STUDIO_YOUR_MODEL_NAME=your-model-uuid-here
  1. Update backend main.py in the /studio/models endpoint:
@app.get("/studio/models")
async def get_studio_models():
    import os
    models = {
        "qwen-gdpr": os.getenv("STUDIO_MODEL_QWEN_GDPR"),
        "granite-gdpr": os.getenv("STUDIO_MODEL_GRANITE_GDPR"),
        "qwen3-sharegpt": os.getenv("STUDIO_QWEN3_SHAREGPT"),
        "your-model-name": os.getenv("STUDIO_YOUR_MODEL_NAME"),  # Add this line
    }
    return {k: v for k, v in models.items() if v}
  1. Restart backend - The new model will appear in the frontend dropdown automatically

Note: The frontend sends raw Studio model UUIDs to the backend (e.g., 7b19cae7-0983-4a6b-a03c). The backend automatically detects UUID format and routes to IONOS Studio API, while provider/name formats (e.g., mistralai/Mistral-Small) route to IONOS Hub.

SetupCI/CD Pipeline