IT Brief New Zealand - Technology news for CIOs & IT decision-makers
New Zealand
Twilio evolves from APIs to conversational trust

Twilio evolves from APIs to conversational trust

Tue, 27th Oct 2026 (Today)
Anthony Caruana
ANTHONY CARUANA Interview Editor

Twilio has evolved over recent years. It was just a few years ago that Twilio called itself "the API company". But it has now altered course using its foundation in bringing communications APIs to enterprises to leap into the world of customer engagement.

Kathryn Murphy, the company's SVP for Product says the went too far towards building applications itself. Twilio now sees its role as providing the infrastructure beneath those applications.

"Five years ago we made a more intentional shift towards engagement, but we also maybe admittedly went too far. We started building apps," Murphy said. "We need to give engagement capabilities but not try to be the app layer. We want to be the infrastructure." 

This evolution has been driven by a shift in customer expectations. Consumers expect that their interactions with an organisation are seamlessly handed over as they move between different functions. Silos, such as those that existed between customer support and marketing, are no longer tolerated. Murphy argued that traditional omnichannel strategies largely added more channels within separate business functions. Marketing might use email and SMS while customer service used voice and chat, but the interactions remained disconnected.

But customers increasingly expect their relationship with a company to persist across functions and channels. This is leading to a shift towards conversations that transcend different channels.

"The reason we're calling it conversation is that it transcends functions. The real driver here is the consumer," she said.

Murphy said consumers increasingly expect communications to be trusted, branded and relevant and no longer tolerate organisations where marketing and customer service operate as disconnected islands. 

For much of its history, Twilio's communications network, that spans the globe through carrier agreements with almost 5,000 carriers across 180 countries, has been a core differentiator between it and potential competitors. However, Murphy said connectivity isn't enough.

"You have to have the reach and the moat that Twilio has in Super Network, but that means nothing if you don't have customer context or if you don't have that data." 

That context is not just about what the customer is asking at that moment.

"I love all these channels, but none of them remember anything," said Murphy.

This has driven Twilio to make the context generated through customer interactions available as persistent conversational memory.

Agent identity

While issues such as context and memory are important, the greatest risk organisations face as the number of AI agents proliferates is security. Once autonomous agents start calling businesses and consumers, communications infrastructure has a new identity problem.

Murphy said there are two sides to this. Businesses need to prove that an agent really represents them, while businesses will also need to establish whether an incoming agent legitimately represents a customer.

"That is the next new problem that Twilio is really worried about: agent identity," said Murphy.

As the number of agent-to-agent interactions increases, the problem will only become greater.

"We always needed to know it was really the consumer, and we always needed to know it was really the business. But now we have a one-step removed problem on both sides of it," she explained.

AI isn't necessarily creating entirely new enterprise security requirements. Instead, autonomy is making weaknesses organisations have tolerated for years much more dangerous. Organisations may have tolerated overly broad permissions for humans, but giving an autonomous agent unrestricted access is a very different proposition. Access control, identity, governance and compliance have always been important, but AI magnifies the risks.

"The agent-to-agent scenario is probably the scariest because there's literally no human in the loop," Murphy said.

Twilio is looking at agent-to-agent communications with governance, guardrails and kill switches. This was a key reason it acquired identity security company Stytch in November 2025.

As AI agents gain greater autonomy, Murphy sees Twilio's role evolving alongside them. The challenge is no longer simply connecting a call, message or application. It is ensuring interactions between people, businesses and autonomous agents remain trusted and governed. That builds on Twilio's existing proposition of providing secure communications infrastructure but raises the stakes considerably. As Murphy put it, AI agents "make this really tricky and cause us to think really differently."