v1.0.1

Release Notes

What's New

White-Label Support

• Made the application configurable for custom branding
• Easy configuration for deploying the software under own brand name

Enhanced Configurability

• YAML-based configuration system for backend and UI settings
• Configurable system messages (e.g., greeting messages, further help messages)
• Flexible model provider support (OpenAI, Google, self-hosted)
• Customizable Redis cache TTL and connection settings

Iframe Integration

• New IframePageInfo model to capture website context
• Bot detects page origin via query parameters (page and title)
• Website context information stored in session state for potential prompt enhancement
• Logger tracks where the bot was called from for debugging

Configuration Examples

See the following example configuration files:

• src/backend_config_example.yaml - Backend configuration
• ui/ui_example_config.yml - UI configuration

v1.0.0

Release Notes (GitHub)

Core Changes

• Model Update

  • Agentic & answer‑generation tasks: now powered by gemini.
  • Law‑level complexity tasks (e.g., summarisation, memory‑management): switched to gemini-2.5-flash-lite.

• Async Refactor

  • Re‑implemented the main components with async/await.
  • All major langgraph / langchain APIs are now async‑compatible, delivering lower latency and better concurrency.

• Docker Optimisation

  • Shrunk the Docker image size (≈ 30 % reduction).
  • Faster build → quicker deployments and lower resource consumption.

• Package Management

  • Replaced pip with uv as the default package manager.

• Dependency Updates

  • Bumped all production‑level packages to their latest stable versions.
  • Note: the streamlit upgrade is still pending and will be delivered in the next minor release.

Performance & Feature Gains

• Increased Scraping Bandwidth

---