Skip to main content

What's New in Version 0.660.0

Released on December 1, 2025

We're excited to announce a major update packed with enhanced AI capabilities, intelligent document discovery, comprehensive analytics, and a refreshed chat experience. This release also marks the private beta launch of Voice to Voice agents! 🚀

🔍 Deep Knowledge Search for Agentic Tooling​

Intelligent Document Discovery​

When you enable Agentic Tooling, Deep Knowledge Search automatically activates to supercharge how your AI agent finds and retrieves information from your Knowledge Hub.

Deep Knowledge Search goes far beyond simple keyword matching. It employs a multi-layered intelligence system that thinks like a researcher, not a search engine.

Fine-Tune Your Search Settings​

The following settings are available when Agentic Tooling is enabled:

  • Min chunks before completeness check - The minimum number of relevant information chunks Deep Knowledge Search must gather before evaluating whether the query has been fully answered. A higher value ensures more thorough research before the completeness check runs.

  • Agentic completeness threshold - The confidence level (0-1) required for Deep Search to consider a query fully answered. A higher threshold means the agent will continue searching until it's more certain the answer is comprehensive. Lower values allow faster responses when partial information is acceptable.

Tip: For quick lookups, use lower values for both settings. For in-depth research tasks, increase the threshold to ensure comprehensive answers. The default values are optimised for most use cases.

📊 Comprehensive Analytics Dashboard​

AI-Generated Performance Insights​

Agents can now track their performance based on customisable instructions against a series of KPIs. The analytics section provides AI-generated insights from chat interactions with filtering by agent and date range, plus CSV export.

Key Metrics Tracked:​

  • Request Completion - Donut chart showing completed vs incomplete chat requests with scores between 0.0 (not completed) and 1.0 (fully completed)

  • Customer Satisfaction - Horizontal bar chart of satisfaction scores across five bands (Very Low, Low, Medium, High, Very High) with average score displayed

  • PII Detection - Tracks personally identifiable information (email, phone, name, address, SSN, credit card) detected in conversations for compliance monitoring

  • Toxicity Analysis - Monitors harmful content including profanity, harassment, hate speech, and threats with average toxicity percentage

  • Escalation Tracking - Identifies conversations requiring human agent intervention with escalation reasons

  • Quality Scores - Distribution of conversation quality ratings (Excellent, Good, Fair, Poor, Unrated) with insights for agent improvement

🧠 Enhanced Sorcerer Models​

Significantly Upgraded Intelligence & Speed​

Both Sorcerer and Sorcerer 1.5 Thinking models have received substantial improvements to their core capabilities. You'll notice:

  • Enhanced Speed: Faster response times across all interactions
  • Improved Intelligence: Better reasoning, more accurate outputs, and enhanced contextual understanding

These improvements make our Sorcerer models more capable than ever for complex tasks requiring deep reasoning and nuanced responses.

💬 Refreshed Chat UI​

Enhanced User Experience​

The chat interface has been refreshed with:

  • More Structured Sections - Especially beneficial when Agentic Tooling is enabled, providing clearer organization of complex interactions
  • Improved Input UX - Better input handling for smoother conversations
  • Enhanced Media Management - Improved UI for uploading and managing media files and attachments

🎙️ Voice to Voice Agents (Private Beta)​

Conversational AI with Real Voice​

We're thrilled to announce the private beta launch of Voice to Voice agents! This groundbreaking capability enables natural voice conversations with your AI agents.

Key Highlights:

  • Out-of-the-box Twilio Integration - Seamless telephony integration for voice interactions
  • More Integrations Coming Soon - Stay tuned for additional voice platform support

Availability: Voice to Voice agents will soon be enabled for Business and Enterprise subscriptions. Contact us to join the private beta!

🔌 API Enhancements​

Raw Stream Mode for Streaming Endpoints​

A new raw_stream parameter is now available for the streaming agent endpoint (/agent on ais.{region}.toothfairyai.com).

How it works:

  • When raw_stream: true, the API prioritises performance by sending token-level SSE events for faster real-time streaming
  • Token events: "type": "token", "message_id": "abc123", "chunk": "Hey"`
  • Final message: "type": "message", ... at stream end

Multilanguage Behaviour: When multilanguage mode is enabled and an intermediate translation pass is required, the stream automatically reverts to sentence-by-sentence delivery to ensure language consistency.

📋 Summary​

This update includes:

  • 1 major model enhancement (Sorcerer models with improved speed and intelligence)
  • 1 intelligent search feature (Deep Knowledge Search for Agentic Tooling)
  • 1 comprehensive analytics dashboard with AI-generated insights
  • 1 chat UI refresh with better structure and media management
  • 1 private beta launch (Voice to Voice agents with Twilio integration)
  • 1 API enhancement (Raw stream mode for improved streaming performance)

Thank you for using ToothFairy! We're constantly working to improve your experience. If you have any feedback or encounter any issues, please don't hesitate to reach out to our support team at support@toothfairyai.com.

Keep building magic with AI! 🧚‍♀️✨

v0.660.0