Twilio Launches Agent-to-Human Protocol
Twilio has introduced A2H, a new protocol for Agent-to-Human communication designed to help AI agents interact with people in nuanced, multi-channel ways. The protocol aims to provide seamless orchestration, error fallback, and cross-channel context. The launch highlights the increasing technical demand on platforms to manage complex interactions between automated agents and human users, requiring robust API design and monitoring.
- The A2H protocol is an open-source specification designed to create a verifiable audit trail when an AI agent requires human approval for an action. It functions as a trust layer, generating a JWS-signed artifact as cryptographic proof that a human authenticated and consented to a specific request, addressing safety concerns with autonomous agents. - This protocol is part of a growing ecosystem of agent-related open standards, including Agent-to-Agent (A2A) for inter-agent communication and the Model Context Protocol (MCP) for providing agents with necessary external information. Twilio is also contributing to extensions for these protocols, such as a latency extension for A2A to help route requests based on an agent's declared response time. - From an architectural standpoint, Twilio's API philosophy centers on providing composable primitives, or "Lego blocks," like webhooks and TwiML verbs, allowing developers to build complex solutions. For AI-driven voice, this is demonstrated in a dual-agent architecture, where a quick "responder" handles low-latency replies while a "thinker" agent manages complex reasoning and context in the background. - The A2H launch aligns with Twilio's broader strategy of embedding AI into its core communication platform, which includes products like ConversationRelay for creating voice AI agents and Conversational Intelligence for analyzing voice and text interactions. These tools are designed to handle massive scale, with the voice platform supporting over 27.9 billion calls a year. - In the logistics and shipping sector, the shift from transactional to AI-enhanced APIs is critical for moving from passive reporting to proactive management. Modern routing APIs can reduce empty miles by up to 25% by analyzing live traffic, vehicle capacity, and driver hours in real time, while API-driven automation in order fulfillment can cut processing time from hours to seconds. - For Q4 2025, Twilio reported revenue of $1.37 billion, a 14% year-over-year increase, with a dollar-based net expansion rate of 109%. Analysts note that the company's Voice AI revenue grew 49% year-over-year, and for fiscal year 2026, the company guided for 8.5% organic growth and over $1.04 billion in free cash flow. - Moody's recently upgraded Twilio's rating to Ba1 from Ba2, citing its strong market position, growth potential from AI, and a financial policy of maintaining cash reserves that exceed debt.
