Would Your Mother Pay Five Pounds? Ralph Page on AI Inside the Operator Network

Ralph Page’s CASA26 keynote on why AI in the conversation belongs inside the operator’s network

Ralph Page has been selling into telecoms long enough to recognise a good era when he sees one. The first came with deregulation, when every new operator wanted the same exchanges as the incumbents, and selling meant asking “would you like one of those?” The second was the PBX death spiral, when cloud providers offered small businesses the features only banks and law firms could afford, and the PBX makers that did not move went under.

Radisys’ Senior Sales Director for EMEA told the CASA26 room he believes he is at the start of his third. And it rests on the same three ingredients as the first two: customer demand, the right technology, and a trusted provider.

People don’t trust AI. They do trust their operator

Cloud telephony took years because businesses did not trust someone else with their numbers and their data. They liked hugging the PBX in the cupboard. Trust came slowly, and then the market tipped.

AI is at the same point, Ralph argued. Ask people to put their details into an AI cloud and many will hesitate. Offer the same capability from the operator that already provides their phone, and they listen. The operators own the conversation. If they bring the AI experience providers into it in a regulated, secure and trusted way, that is what drives the business.

Radisys has spent two decades doing media processing inside operators’ IMS networks, and built a CPaaS platform on top of it years ago. Selling that platform was hard. Product managers understood it but found it too complex to take to market. AI changed that: integrate once into the network, and AI can be added to calls at the scale operators are used to.

From two years to seven months

The change shows in the timelines. Ralph’s first deployment, with a large US operator, went from the first calls in the lab to the edge of commercial launch in about two years, which anyone who sells to tier-one operators will recognise. The second, in Japan, went from order to production in around seven months. In both, the lead application was call recording with transcription and summaries, used to get the platform into the network before adding more.

The buyers have changed too. Two years ago Ralph talked to operations teams who might push an idea upwards. Now senior executives, and sometimes board members, approach Radisys themselves, and the RFPs describe real products with an idea of how they will be priced and sold next year.

The five-pound test

On pricing, Ralph sees two models: a monthly fee per user, or a bundle. And he offered a simple test for which services will sell. His 93-year-old mother is terrified of fraud and barely answers the phone. Would she pay five pounds a month for an AI that listens in the background and whispers a warning when it suspects a scam or a deepfake voice? She would bite your hand off. His father struggles to hear the doctor on the phone; he would pay for clarity too.

A digital assistant for Ralph himself? Probably not. For a plumber who misses one call a week from a customer who would need a boiler serviced for ten years, an assistant that answers, checks the diary and books the job is worth far more than five pounds. The value is not the AI. It is the problem it removes.

Regulation is the operator’s advantage

Radisys debated for years whether to offer its platform as a hosted service. It has decided to deliver it into the service providers, because they are the ones who can handle the regulation. Ralph’s example: under EU rules, everyone on a call must agree before AI is used. So if you want live translation with someone who speaks another language, how do you get their consent without using AI? Radisys uses keyword detection, which is not AI, to hear the request, pauses the call, asks the other party in their own language, and records the consent. Operators know how to do that. Most AI experience providers do not want to.

The other lesson is speed. One customer building live translation is on its third choice of AI model in eighteen months, because the technology moved from turn-by-turn translation to real conversation. Operators need an orchestration layer that keeps them current. Radisys’ answer, announced that week, is V.AI, an ecosystem of AI partners it can bring into operator deployments, including ElevenLabs and Speechmatics.

Fraud alerts, hearing support and assistants are what telecom people think of first. What excites Ralph more is everything the vertical experience providers will build once the network is ready: for dentists, doctors, lawyers and receptionists. The operator brings the network and the regulation. Others bring the ideas.

Our thanks to Ralph and to Radisys, a long-standing CPaaSAA member and CASA26 partner, for a keynote that answered the questions we asked: real deployments, real pricing, and what the last two years taught them.

More from CASA26

CPaaSAA members have every CASA26 session in full in the member community.

Not a member yet? Tell us you are interested and we will be in touch.

Following the series? New sessions appear here every week until Mobile World Congress. Subscribe on YouTube to catch each one.

CASA26 ran 21–23 September 2026 at the Tobacco Theater in Amsterdam, hosted by CPaaSAA. This is part of our Tuesday series on Intelligence.