Editorial illustration for Twilio wants AI agents to access customer communications directly
Twilio Deploys AI Agents to Handle Customer Communications
Twilio's chief operating officer Thomas Wyatt has a straightforward pitch for what AI should actually do inside a contact center: handle the boring stuff. In an exclusive interview with AIM, Wyatt argued that enterprises are wrong to frame AI as a wholesale replacement for human customer service teams. The better model, he says, is a split, where AI agents absorb level-one and level-two queries, the repetitive password resets and billing questions that eat up staff time, while human agents get pushed toward the harder cases that need judgment or specialized knowledge.
That shift changes what the job of a human contact center worker looks like. Instead of fielding everything that comes in, staff increasingly deal with the fraction of cases AI cannot close out, the ones that get escalated because something about them resists a scripted resolution. Wyatt says Twilio has built tools like Conversation Memory specifically for that handoff moment, so a customer who starts with a bot and ends up with a person does not have to explain the problem twice. Whether customers themselves will start picking AI over humans, and what that means for how companies staff support lines, is the question Wyatt turns to next.
AI agents are taking over routine customer service tasks, pushing human contact centre workers towards more complex problems that require judgement, specialised knowledge, and human intervention.
Why this matters
Twilio's pitch is really an infrastructure land grab. If AI agents get direct access to the same communication rails that already handle SMS, voice, and chat for thousands of enterprises, Twilio becomes the default plumbing for agentic customer service rather than just another vendor competing for attention. For developers and founders building on top of these APIs, that's worth watching closely: whoever controls the channel layer gets to set the terms for how agents authenticate, hand off to humans, and log interactions for compliance.
Wyatt's framing, that AI should absorb level-one and level-two queries while humans handle judgement calls, sounds reasonable, but it also conveniently justifies Twilio selling more infrastructure rather than fewer seats. The real test isn't whether AI agents can access communications APIs; plenty of companies can build that. It's whether Twilio's existing enterprise relationships let it move faster than newer agent-platform startups chasing the same contact-center budgets.
Anyone building customer-service agents right now should be asking who owns the handoff logic between bot and human, because that's where the actual product value, and the lock-in, will live.
Related Reading
Industry Applications Westinghouse teams with Google Cloud to build AI platform for nuclear power
Industry Applications NVIDIA NeMo powers telco reasoning model for autonomous network workflows
Industry Applications Month-1 Agent Adds Holistic Observability with Trace IDs and Token Tracking
From Archives Early AI Adoption by Industry Shows What Other Sectors Can Learn
From Archives Europe’s Industrial Leaders Use NVIDIA for Up to 40x Engineering Speedups
Common Questions Answered
What is Twilio's vision for how AI agents should be used in contact centers?
According to Twilio's COO Thomas Wyatt, AI agents should handle routine, repetitive tasks like level-one and level-two queries, password resets, and billing questions rather than replacing human customer service teams entirely. This split model allows human agents to focus on complex problems that require judgment, specialized knowledge, and human intervention, creating a more efficient division of labor.
How does Twilio plan to position itself as the infrastructure provider for AI agents?
Twilio aims to give AI agents direct access to the same communication rails that currently handle SMS, voice, and chat for thousands of enterprises. By controlling this channel layer, Twilio becomes the default infrastructure for agentic customer service rather than just another competing vendor, allowing the company to set terms for how agents authenticate and operate.
What types of customer service tasks are AI agents best suited to handle according to the article?
AI agents are best suited for handling routine and repetitive customer service tasks, specifically level-one and level-two queries such as password resets and billing questions. These are the types of inquiries that consume significant staff time but do not require the judgment, specialized knowledge, or human intervention needed for more complex customer issues.
Why should developers and founders building on Twilio's APIs pay attention to this strategy?
Developers and founders should monitor Twilio's infrastructure play closely because whoever controls the communication channel layer for AI agents will have significant power to set the terms for authentication, integration, and operation. This control could determine the competitive landscape for AI-powered customer service solutions built on top of Twilio's platform.
Further Reading
- Twilio revamps, updates customer engagement platform - TechTarget
- Infrastructure for the agentic era: A new conversation layer for the Twilio Platform - Twilio
- Twilio's Next Generation Platform: An Infrastructure Layer for Every Conversation - Twilio
- 7 top customer data platforms for AI agents in 2026 - Twilio
- 8 top AI-powered contact center platforms in 2026 - Twilio
