Twilio Tackles 'Agent Amnesia' With Persistent Memory for AI Agents
Twilio launched four new capabilities on May 6 designed to give AI agents persistent memory across customer interactions, so customers stop repeating themselves when switching channels or getting transferred.
The problem is real. When a customer calls support, explains their issue, gets transferred, and starts over-that friction multiplies with AI agents in the mix. Twilio calls this the "conversation gap": the disconnect between fragmented systems that leaves agents without context.
Most businesses don't recognize the problem, according to Omar Paul, Twilio's vice president of product management. "Most businesses are operating in the dark," he said. "They have siloed data sitting in different systems inside that enterprise. They struggle with fragmented channels, and the integration complexity that is required to link these things together is at a breaking point."
Without shared memory of previous interactions, agents-human or AI-start blind every time. AI was supposed to be the easy button for customer engagement, Paul said, but "this reality has not lived up to the hype."
Four New Capabilities
Conversation Memory maintains history, preferences, behavior, and conversation state in real time across every interaction and channel. When a customer contacts a brand, the agent already knows who they are and where things left off.
Conversation Orchestrator connects individual calls and messages across channels into one unbroken conversation. It manages routing, escalation, state, and handoffs between humans and AI so a customer never feels like they're starting fresh because they switched from chat to phone.
Conversation Intelligence watches interactions as they happen and triggers automated responses in real time. It detects when a conversation is going sideways-say, an AI agent struggling with a customer-and escalates to a human. Paul called this "a relevant but surprisingly difficult thing to do" for most businesses today.
Agent Connect is a self-hosted open-source framework that lets companies plug their own AI agents and models directly into Twilio's voice and messaging channels without rewiring their existing stack. Businesses keep their preferred AI and models while Twilio handles the underlying complexity of real-time voice streaming, session management, and identity.
The Real Competition Is Infrastructure
Salesforce, Amazon Web Services, and Zendesk are all moving into this space with their own context-aware customer engagement tools. But Twilio argues the battle won't be won at the feature level. The real advantage lies deeper-at the infrastructure layer, where conversations are actually built, routed, and remembered.
Mila D'Antonio, principal analyst for customer engagement at Omdia, said in a statement: "At the center of the CPaaS, CDP, and AI convergence, Twilio is redefining what a Customer Engagement Platform looks like-one that remembers, adapts, and orchestrates across every touchpoint."
All four capabilities are available starting today. For teams managing customer support, these tools address a core frustration: customers shouldn't have to repeat themselves. Learn more about AI for Customer Support and how AI Agents & Automation are reshaping how businesses handle customer interactions.
Your membership also unlocks:
