Voice-enabled conversations and advanced content processing.
Voice-Powered AI Chat
Voice support now extends Private Data AI Chat and Public AI Chat. Visitors and internal users can ask questions out loud, listen to answers (public chat only, private chat responses are often complex and not suitable for voice), and move between voice and text without losing context in the same conversation.
Voice input and output Ask questions verbally and receive audio responses read aloud for hands-free interaction. Voice runs across web, mobile, and desktop experiences.
Enhanced chat capabilities Voice recording and playback live directly in the chat interface, with higher audio quality and more reliable streaming. Partial responses arrive as they are generated so longer answers do not block the interface.
Multi-model voice intelligence Access leading AI models from OpenAI, Anthropic, Google Gemini, xAI, and Cohere for voice interactions. Intelligent model selection balances reasoning quality, latency, and cost for each conversation.
Voice-enabled public chat Website visitors can now use voice in the same Public AI Chat widget that already handles text questions and lead capture. Voice interactions can lead into sharing contact details or booking time with your team when configured.
Accessibility and usability Voice capabilities improve accessibility for people who prefer listening over reading or use assistive technologies. The experience is tuned to work reliably across devices and varying network conditions.
Advanced Content Processing
Output from AI chat and AI Workflows is now easier to adapt into downstream systems, templates, and reports. December adds improvements to structured content handling, version comparison, and markdown rendering across the Clear Ideas interface.
Handlebars variable support Enhanced template processing with syntax highlighting and variable recognition for more sophisticated content automation.
Version comparison tools Compare different versions of extracted text content to track what changed and why. Diff-aware views help review AI outputs, regulatory updates, or content revisions over time. See Viewing Files for more details.
Enhanced markdown processing More robust markdown handling and link processing ensure that complex responses render cleanly, whether they are viewed in chat, exported, or reused inside other experiences.
Public Engagement Features
Extended reach and interaction:
Voice-enabled public chat Website visitors can now engage with AI chat using voice, expanding accessibility and user experience options.
Lead capture integration Automated email capture and follow-up systems to nurture visitor engagement and conversion.
Site Management and Safety
Better control and protection:
Deletion protection safeguards Added safety measures to prevent accidental site deletion, with clear confirmation processes and recovery options. Enhanced site management features with improved error handling and user cleanup logic.
Enhanced site management features Improved billing features and organization policy enforcement with streamlined permission checks and clearer error messaging. Better organization membership verification and site access controls.
Advanced site configuration Enhanced site settings and menu item access controls with improved user selection logic and notification management. Refined site role handling with better user experience and error clarity.
Developer and Integration Tools
Enhanced connectivity and control:
Advanced MCP capabilities Expanded Model Context Protocol support with content listing, diff retrieval, and improved access management. Secure integration allowing compatible AI applications to query private data with controlled access and full audit logging.
MCP file and folder operations New parameters and shared helpers for advanced file management through Model Context Protocol. Enhanced scope management ensures only authorized applications can access specific content areas.
MCP access control with refined scope Granular permission controls for MCP connections, allowing fine-tuned access to specific sites, folders, or content types while maintaining security boundaries.
MCP tools with content listing and diff retrieval Advanced MCP functionality for listing content, retrieving differences between versions, and performing sophisticated content operations through secure API connections.
Refined error handling Clearer error messages and better feedback throughout the platform for improved user experience. Enhanced logging and debugging capabilities for troubleshooting integrations.
Enhanced AI model support Improved AI model configurations with error handling in streaming responses and enhanced integration capabilities across different AI providers.
AI Workflow Builder Enhancements
Advanced automation creation and refinement:
Intelligent document analysis Automatic document type detection and complexity assessment for optimal workflow generation. Smart variable discovery and pattern extraction from document structures.
Multi-stage build process Sophisticated five-stage workflow creation including analysis, generation, benchmarking, revision, and finalization. Real-time build progress monitoring with detailed status updates.
Quality benchmarking and iteration Built-in quality assessment with automatic prompt refinement. Iterative improvement system that learns from failures and creates better prompts through multiple optimization cycles.
Document structure preservation Maintains logical flow and structure of original documents while making them dynamic and reusable. Preserves formatting, section relationships, and content hierarchy.
Cost estimation and token management Advanced cost prediction and token usage estimation before workflow building. Optimize for efficiency and budget management with transparent pricing.
Multi-Step Loop Processing
Powerful branching and iteration capabilities for complex data processing in AI Workflows:
Container steps with child steps Create container steps that execute multiple sequential child steps for each iteration. Process collections of data through multi-step pipelines where each child step builds on previous results.
Enhanced variable scoping Hierarchical variable access across workflow, step, and loop contexts. Child steps can reference parent step outputs, loop variables, and previous child step results using nested references like
{{step-2-1}}.Visual workflow editor improvements Clear visual representation of container steps and their child steps with intuitive connection indicators. Improved step reference detection and unused variable warnings help identify potential issues before execution.
Variable insertion tools Quick variable insertion dropdown in prompt editors with syntax highlighting. System variables and defined variables are clearly distinguished, making it easier to build complex workflows without manual typing.
Advanced loop options Flexible loop configuration with customizable item and index variable names. Conditional filtering using comparison and logical operators to process only relevant items from collections.
Improved output aggregation Automatic JSON array aggregation for container steps with multiple iterations. Final outputs combine results from all iterations while preserving structure and enabling downstream processing.
Enhanced Analytics and Search
Improved discovery and insights with Analytics and AI Enhanced Search:
Advanced search capabilities Enhanced search result filtering with flexible date range presets and sticky filter settings. Improved search query handling and result ranking for better relevance.
Analytics enhancements Time-series and list views for engagement metrics. Open items directly from analytics for deeper investigation.
Under the Hood
Quiet improvements that make everything feel steadier: enhanced connection handling, improved content sanitization, and optimized processing pipelines across all services.