Twilio Targets AI-Powered Customer Conversations With New Orchestration Tools

Twilio Targets AI-Powered Customer Conversations With New Orchestration Tools

Twilio is expanding beyond communications connectivity with Conversation Memory, Conversation Orchestrator and Conversation Intelligence, aiming to help businesses coordinate AI agents, human agents and multichannel customer interactions. Voice AI adoption remains early because accuracy, latency, voice quality, turn detection, trust and regulation remain challenges. Twilio plans to remain model-neutral, allowing customers to use multiple AI, speech-to-text and text-to-speech providers. Twilio reported broad-based organic revenue outperformance, with messaging growing about 18% in the first half and voice revenue rising more than 20% in the second quarter. Higher-margin voice and software add-ons, along with cost reductions, are supporting gross-profit growth.

Twilio NYSE: TWLO executives outlined the company’s strategy to expand beyond communications connectivity into tools designed to provide context, orchestration and intelligence for interactions involving customers, human agents and artificial intelligence systems. Speaking at a Goldman Sachs event, Twilio said its core business remains connecting customers with end users through communications channels. However, the company sees its newer conversation-focused products as an important part of its future, particularly as businesses deploy AI agents alongside human support teams. Chief Product and Technology Officer Inbal Shani described Twilio’s platform as consisting of three layers: communications channels, contextual data and AI agents operating across those channels. The goal, she said, is to use real-time context to make AI agents “more effective, more productive, more accurate.”

Conversation products and developer flexibility Twilio recently launched products including Conversation Memory, Conversation Orchestrator and Conversation Intelligence. Shani said the company is seeking to preserve its developer-first approach while also making it easier for a broader set of users to build customized solutions. “The concept of developer is changing,” Shani said, noting that declining development costs are enabling more enterprises, independent software vendors and AI-native companies to create tailored applications. Conversation Memory is intended to help preserve context across customer interactions. Shani said Twilio is distinguishing between information needed to improve a real-time conversation and longer-term data held in systems such as customer relationship management platforms and data warehouses. Rather than asking customers to duplicate their existing data, Twilio is building connectors to those systems and retaining information most relevant to the interaction. Beta customers helped shape product priorities, according to Shani. One key request was a “warm handoff” between an AI agent and a human agent, as well as the ability to detect when an interaction should be escalated. While Twilio initially emphasized customer-support applications, some beta users also adopted the products for sales uses, such as identifying leads outside business hours and transferring them to sales staff later.

Voice AI opportunity remains early Twilio said voice AI remains in the early stages of adoption, with challenges involving latency, quality, voice quality, turn detection, background noise, network variability and model accuracy still being addressed across the industry. Shani said accuracy is the primary barrier to deploying voice AI agents at scale, and that infrastructure is especially important for managing late...