Product changelog and announcements

Conversation Relay Insights provides the observability to confidently build and scale your Voice AI agents. By analyzing end-to-end conversational signals—like latency, interruptions, and handling time—the dashboard highlights exactly where your agents excel at natural turn-taking and where friction may be emerging. This visibility allows teams to replicate high-performing interaction patterns, validate deployments, and systematically deliver more natural conversations.

For deep-dive troubleshooting, the call summary view allows you to inspect individual interactions with a synchronized event timeline and recording playback with visualized audio waveforms. This level of detail enables developers to reconstruct the exact flow of a conversation, pinpointing the bottlenecks that impact your customer's experience.

These features are available exclusively as part of Voice Insights Advanced Features.

Learn more in our Conversation Relay Insights documentation.

Conversation Intelligence is now generally available with real-time, cross-channel support. Conversation Intelligence transforms customer interactions into actionable signals — in real time or post-conversation — across channels. Part of the Twilio Conversations layer, it integrates natively with Conversation Orchestrator, Conversation Memory, and Enterprise Knowledge.

This release includes:

• Real-time intelligence: Supports live agent assist, real-time automation, and AI agent observability use cases

• GenAI-native Language Operators: four Twilio-authored (Next Best Response, Script Adherence, Sentiment, Summary) plus fully custom operators powered by LLMs

• Cross-channel support: One unified intelligence layer across Voice, SMS, MMS, RCS, WhatsApp, and web chat.

• Business & customer awareness: Enrich operators with Conversation Memory and Enterprise Knowledge

• Composable developer primitives: Create Rules combining triggers, operators, context, and webhook actions — all managed via REST API

• Insights: Aggregate operator results across conversations for business-wide patterns

Read the announcement blog post for the full story, or get started with the Conversation Intelligence docs.

Note: The previous version of Conversation Intelligence (post-conversation, voice only) will now be referred to as Conversation Intelligence (classic) across Twilio's assets.

Twilio Agent Connect is now generally available with multi-channel support. Agent Connect is a Python and TypeScript SDK that simplifies building AI-powered conversational agents with persistent memory and conversation orchestration—connecting your LLM application to Twilio's communication channels.

This release includes:

• Multi-channel orchestration: Built-in support for Voice, SMS, chat, WhatsApp, and RCS with unified callbacks

• Conversation Orchestrator integration: Automatic conversation tracking, participant management, and cross-channel support

• Memory integration: Automatic customer memory retrieval with semantic recall, profiles, and conversation history

• Conversation lifecycle management: Session tracking, profile management, and conversation-ended hooks across all channels

• LLM adapters: OpenAI SDK adapter with automatic memory context injection for faster development

• Built-in tools: Pre-built tools for memory retrieval, knowledge search, and Flex escalation compatible with OpenAI and Anthropic function-calling

• Cloud platform integrations: Azure (Agent Framework, Voice Live) and AWS (Strands, Bedrock, AgentCore) connectors for cloud-native deployments

• Production-ready server: TACServer with FastAPI (Python) or Fastify (TypeScript) handles WebSocket and webhook endpoints automatically

 

Read the blog post for the full story, or get started with the documentation.

 

Note: Agent Connect supports Conversation Orchestrator for multi-channel applications (recommended), or ConversationRelay-only mode for voice-first use cases. Optionally integrates with Conversation Memory and Conversation Intelligence for enhanced capabilities.

 

Twilio Conversation Memory helps AI and human agents remember key facts, past conversations, and customer preferences so they can deliver more personalized responses. By feeding the AI agent only relevant context, Conversation Memory helps maintain agent performance while lowering token usage and response latency. As a result, every conversation can build on the last instead of starting from scratch.

Twilio Enterprise Knowledge is generally available. It enables agents to pull business context into conversations from unstructured sources such as knowledge bases, company policies, and documentation, helping them answer customer questions with greater accuracy and relevance.

Enhance your voice agents with Deepgram’s Flux model for better turn-taking and fewer interruptions. This update also introduces new TwiML attributes for precise conversation flow control and a dedicated suite of error codes for easier debugging.

TLDR: We are launching Voice Recording Configurations; resources for defining, applying, and managing call recording settings at scale. Voice Recording Configurations are available in GA in the us1 region.

• Simplify sending and build faster with a single shared API that is compatible with Android, iOS and web browsers 
• Automatically expire stale tokens instead of managing expiration webhooks and token state
• Send up to 10,000 notifications with a single API request and personalize those notifications for each recipient using Liquid
• Schedule notifications for a later time

This API also features shared primitives that make it easy to extend your use case across SMS/MMS, RCS and Email using the Bulk Messaging API and Twilio Email. Request access to the private beta to try it today!

Note that this API is expected to replace Twilio Notify. If you are a Twilio Notify customer, please indicate it in your request for faster access so you can start planning your migration.

The Bulk Messaging API is now in Public Beta, enabling product and engineering teams to message at scale. Key features and benefits include:

• Send up to 10,000 messages with a single API request over SMS/MMS and RCS, eliminating the need for custom queuing, rate limiting, and retry logic
  
• Personalize message text, images, rich content, and tags for each recipient using Liquid 
• Schedule bulk sends for a later time
• Send RCS rich content without templates, or continue using your existing Content Templates

This API also offers full JSON support, consistent versioning schemes, and shared primitives that work across Twilio Email and Push Notifications, enabling you to expand to multiple channels with fewer lines of code.

By providing multi-recipient sends, per-request controls and a durable API design, the Bulk Messaging API reduces implementation complexity and can accelerate time-to-scale for teams of all sizes. Check out our tutorial to get started now.