Trillet Change Log

Major platform upgrade with custom SIP, workflow automation (n8n, make.com), advanced permissions, and enhanced call management for a better enterprise experience.

Custom SIP Architecture integration for enhanced telephony infrastructure.

In-call API field mapping with dynamic key-value pairs for advanced post-call analysis variable extraction.

New API handlers in documentation for comprehensive workspace management capabilities.

n8n and make.com integration now available for seamless workflow automation.

Dispositions v2 with enhanced call outcome tracking and analysis features.

Added quick action to email support.

Prior calls context upgrade: now keeps track of dynamic variables captured over the last 5 calls, allowing call resumption from where they left off if a caller rings back or is busy.

Permissions implementation for granular access control and security.

Text streaming issue resolved for improved real-time communication.

Batch calls error display issue resolved for better error visibility and debugging.

Auto recharge button not getting disabled after being turned off—now properly responds to user settings.

Resolved issue preventing duplicate workspace creation from multiple clicks.

Responsiveness issues resolved across the platform for improved UX on all device sizes.

Warm Transfers enhancement with configurable timeout and timeout message for better call handling and return to caller.

Implemented a brand new UI design for a fresh, modern user experience.

Redesigned the agent creation experience for a more intuitive and streamlined workflow.

Added the ability to redirect return inbound calls to a specific phone number.

Introduced MCP integration, allowing agents to connect to any MCP server for seamless integrations.

Addressed various issues with batch calls for improved reliability.

Made significant improvements to smart call functionality for better performance.

Enabled exporting of call history data, including full transcripts and enriched metadata.

Launched Batch Call APIs to programmatically manage call queues. See docs at docs.trillet.ai.

Implemented a 2-minute timeout for calls where the caller remains silent, automatically ending the call.

We’re excited to announce the deprecation of Advanced Call Flows in favor of our revolutionary new Visualizer tool, designed to transform how you create and optimize AI agents.

Introducing the Visualizer: A powerful new tool for visualizing, editing, and fine-tuning agent prompts with intuitive loop detection and optimization capabilities.

Enhanced prompt visualization with real-time feedback and suggestions for improved agent performance.

Advanced loop detection and optimization tools to help create more efficient and natural-sounding conversations.

Streamlined UI for better workflow efficiency and user experience.

Improved error handling and validation for more reliable agent creation.

Various bug fixes and stability improvements across the platform.

Major agentic voice AI infrastructure upgrade for improved latency, better tool calls, enhanced reliability, expanded voice detection model options & much more.

Added Gemini 2.0 Flash model with 1M context window and production-ready features back.

Introducing Prompt IDE for call flows: visualize tools such as calendar functions, In-Call APIs and transfers within the prompt editor, with autocompletion and instant error detection.

Implemented a prompt versioning system, maintaining 5 versions, version differences, and notes allowing you to easily switch between different prompt versions, instantly visualize differences and increase productivity.

Organize your call and message flows into folders for better structure and efficiency.

New prompt overwrite warning when saving call flows to support cross-team collaborations.

Enhanced logging for calendar and API tool calls, providing detailed error feedback and payload views.

Platform-wide visual loading cues improved for a smoother user experience.

Advanced search, category filters, and default grid view for call/message flows added for better flow management.

Added ability for agents to use an interactive ‘thinking’ sounds to signify processing states - for better customer experience.

Introduced a prompt field for dynamic AI welcome greetings for granular control and improved latency.

Omni-Channel Inbox: Introduced message and conversation deletion capabilities, plus customer renaming option.

Persistent volume controls during testing, maintaining user preferences across sessions.

General UI/UX optimizations for enhanced interaction.

New Spanish voice Isa added for expanded linguistic capabilities.

Accurate field names appear in batch calls editor - as in original CSV.

Restrictions applied on ‘GET’ method for In-Call APIs, removing objects & nesting.

Resolved persistent ElevenLabs API key issues when switching from BYO mode.

Removed preview text modal during voice selection for instant previews with preset text.

Multiple bugs fixed.

Speed, similarity, and stability enhancements for faster, more accurate, and reliable interactions

Improved live logging while testing on web for better visibility and debugging

Agents now fully initialize before calls connect to avoid startup delays

Improved handling of call endings for smoother call conclusions

More reliable and seamless call transfers

Call histories now show more comprehensive and visible interaction data

More accurate and insightful post-call analysis

Persistent dynamic variables for testing across sessions

Beta support for GPT-4.1 and GPT-4.1 Mini

Interactive Voice Response (IVR) system

Custom greetings for inbound and outbound calls; dynamic greetings are now call-direction aware

Pricing page is now more customizable

UI enhancements

Bug fixes

Google Sheets integration

Nested variables for incall API

UI enhancements

Batch Call improvements

Agent enhancements

Improvements to cal.com integration custom variables

Bug fixes

Over the coming weeks we will be prioritizing refining all current features and reliability will be our number 1 priority.

Text to speak during cal.com API calls

Nested objects in realtime API

Custom data in native cal.com integration

Export call histories

Accurate English STT (a bit slow)

Better pronunciation for Rime / Trillet voices (using brackets)

Batch call recreation feature - still ongoing

Multiple bug fixes

Various UI enhancements

Introduced Trillet Voices for enhanced speech quality and naturalness

Implemented auto flow generator for simple flows to streamline call flow creation process

Improved Knowledge Bases with better retrieval and integration capabilities

Enhanced whitelabel functionality with expanded customization options

Updated Pricing Page with clearer tiers and feature comparisons

Improved post call webhook to include room name in call metadata

Simplified Agent Creation Flow for better user experience

Enhanced Autoflow Generator with more intelligent conversation path creation

Various UI/UX improvements across the platform

Multiple bug fixes and stability improvements

Implemented intelligent STT provider selection between Speechmatics and Deepgram based on language for optimal accuracy

Added custom domain support for Web SDK integration on self-hosted websites

Introduced web call recording functionality

Added call flow import/export capability for easy workflow sharing

Expanded supported languages list with improved regional options

Enhanced Eleven Labs custom voice integration user experience

Resolved batch calling issues for UK, Canada, and US phone numbers

Various bug fixes and UI improvements for better usability

Introduced Simple Prompt Templates with best practices to streamline prompt creation

Added Human-Like Tone toggle for more natural voice responses

Enhanced AutoBuilder with first-speaker selection for optimized call flow generation

Implemented comprehensive UI improvements and validation fixes

Call Flows now have Warm and Cold transfers

Beta call analytics dashboard with more coming shortly

Dynamic Variable replacement when testing web calls or sending single calls

cal.com integration natively

BYO elevenlabs, bring your own custom trained voices to elevenlabs today!

Call Flows further improvements

End Call improvements

Improvement for API speed when fetching call history

External Numbers fix, bringing numbers via SIP

Message flow creating agent fix

Active sessions close fix for calls

Major call flow improvements with enhanced node visualization and real-time state tracking

Enhanced end call detection with improved goodbye message handling

Updated minimize/expand toggle with intuitive directional arrows

Optimized Agent Flow Visualizer layout with compact view mode

Custom Domain for White Label customers

New recording Modal and public recording URL on history page

Nova 3 transcription engine with improved accuracy for Names, Numbers and general data capture

Call flow upgrades lower latency between nodes and end call upgrades

Updated supported voices

Improved Voice Mail detection

Added search and filter functionality for call logs

Fix for phone number not removing on phone management page

Released Call Flows and Editor in Alpha with enhanced node transitions for seamless long or advanced conversations with automatic variable extraction

Added Rime AI integration as a new Text-to-Speech provider

Implemented Security Opt-out feature with automatic transcript and data removal upon webhook trigger

Redesigned homepage to align with new branding

Introduced new settings menu in call flows for improved user experience

Added configurable reminder settings and silence-based call termination for agents

Enhanced AI agents with improved voice detection and variable capture capabilities

Various improvements and bug fixes for web call SDK

Agency mode, now has the ability to add branding to the master workspace

Added Gemini 2.0 Flash Lite model with 1M context window and ultra-fast responses

Added Gemini 2.0 Flash model with 1M context window and production-ready features

Updated call settings with silence handling - includes reminder messages for silent callers and automatic call termination for non-progressing calls

Enhanced Gemini models with more friendly conversational style for improved user interactions

Fixed white label bug causing incorrect branding display

Fixed Stripe bug when managing subscriptions to ensure accurate billing updates

Fixed various small UI issues for improved user experience

Added public and private agents for web SDK

Added reanalyze button functionality to replay call analysis and trigger new webhooks

Added configurable recording URL visibility with public and private options

Added non-interruptible option for both custom and AI-generated welcome messages

Added Krisp noise cancellation for improved audio quality in both phone and web calls

Added role-based access control for wallet sharing options

Integrated phone number subscriptions with Stripe payment system

Resolved billing issue where sub-workspace charges were incorrectly applied to master wallet

Implemented various UI improvements and bug fixes across the platform

Introduced changelog system to keep users informed of platform updates

Improved call transfer system with enhanced cold and warm transfer capabilities

Optimized background noise loading to reduce latency in call initiation

Optimized agent data management by implementing selective data loading

Migrated auto recharge system to Stripe for improved payment processing

Enhanced call conclusion with improved farewell message system

Fixed missing breath pauses between sentences for specific voice models