RCS: Why the new messaging standard could catch on thanks to Apple

Twilio was founded in 2008 to combine communication channels on a software-as-a-service platform. The US company became known to normal users as a service provider for Signal and the parent company of Authy, among other things, but countless other companies use the platform. In this interview, Head of Product Kathryn Murphy explains what the company does and how important Apple's decision to finally join RCS is.

heise online: Twilio is one of those big players that only specialists take notice of, even though your company is now an important part of the Internet infrastructure. You manage the communication channels for many different companies. How does Twilio fit in with other cloud platforms such as Google Cloud and Amazon Web Services?

Kathryn Murphy: I think a lot of our customers use us in addition to those platforms because we are a different type of infrastructure. Think of us as a last mile communications provider. You might be running large applications or services on AWS, but when you finally want to deliver a message to your customers, whether it's via email, SMS, WhatsApp or some other service, that's on a different track, so to speak. And that's where we come in.

How do you think communication will change?

I would say that the trend is generally moving towards direct conversations between customers and companies rather than simple notifications. We are seeing more and more direct communication taking place – or at least attempts to do so. We see this in our own data.

For example, we tracked various WhatsApp messages that were supposed to be simple notifications. And we found that 33 percent of consumers replied anyway, even though they were told that no one was reading them. Even when we send simple one-time passwords, which is a key use case of our platform that we offer, five percent of people still respond. So one of the big changes we're seeing is that consumers increasingly expect to be able to respond to companies on any channel, whether it's an inbound company that customers are approaching or an outbound company that is marketing directly.

So companies should make sure that there is always someone on the other side?

Well, that should at least be the case. The simple fact that we know that these communication attempts are taking place is already very helpful. I think most companies only have a certain way of communicating with their customers, and we try to help them evolve so that communication works everywhere.

Ultimately, it's about dialog with customers. They may start with a simple marketing message, but when people respond with a question, you want to direct them to a customer service representative. Twilio provides a platform for that. We have all these channels.

Are there differences between the communication channels you offer? The content of a simple notification must be different from that of an e-mail.

When customers come to Twilio, they can choose which channels they want to use and they have full control over which channels they use. However, we are well aware that it is becoming more and more annoying for consumers, they are flooded with information and notifications.

We are trying to counteract this by making it easier for companies to take customers' preferences into account and only consider approved channels. So if the customer says they don't want to hear from you, they won't if nothing goes wrong.

Speaking of channels that people no longer want: When will SMS, a notoriously completely outdated technology, finally die?

It will take time. Perhaps at some point SMS will only be used from person to person on feature phones and less from company to customer. But I think it's still a valid channel, unfortunately.

Network operators have been trying for a long time to get Rich Communications Services (RCS) off the ground to replace SMS with modern features.

We have literally been working on RCS for four years. Now it's finally here and we're very excited about it. We believe this is the future of messaging for businesses. The problem is that it could be two, three or even four years before it's fully established. So SMS will have its place on our cell phones for a while yet.

When integrating RCS, we have therefore also created fallback options. Even if our platform assumes that you have an RCS-enabled phone, if we can't deliver the RCS message, we will automatically try a normal SMS.

Even on the Android side, there seem to be problems, even though Google has been pushing the issue for a long time.

There are simply so many device variants on the network that it will take some time for the system to establish itself on a broad front. But it's all very exciting and the potential is huge.

And, of course, people have to want to switch to RCS.

I think as we get more and more RCS messages, many of us will say: "This is much better. Why should I accept any other means of communication? I'm not going to bother with it anymore." And I think companies will have to move quickly in that direction.

Are you satisfied with the current functionality of RCS? It's not fully encrypted, for example, so on the iPhone, where it can now finally be used, people tend to stick to Apple's iMessage.

I think we're off to a good start here and it will get better quickly.

Apple caused a stir when it finally announced RCS support after years of pressure from Google and others.

With Apple's announcement of RCS support, the service is becoming more and more of a winner as it is not platform specific and companies don't have to implement so many different ways to communicate with their customers. We'll see what happens with iMessage.

RCS has been around for a long time, but has hardly been used. So did Apple provide the decisive impetus?

Apple has a big part in it because of its influence. But the other part is that it took so long for the network operators to jump on the bandwagon. And although RCS has been around for a long time, network operators worldwide have only jumped on the bandwagon in the last twelve months or so. And that is the current state of affairs.

Nothing seems to work without generative AI. Twilio is working on virtual call center agents in which AI systems take over telephone calls. If this really works technically, why is it such an interesting market?

One of the things we often forget is that a virtual call center agent has unlimited time. Many call centers are under pressure to close calls quickly, even with customers that are valuable to them. But with a virtual agent, there is no limit to how many calls they can complete.

We have some clients with physical stores whose store managers have to take calls and also receive online orders. The companies put pressure on these managers to complete the call in less than 90 seconds. In comparison, a virtual call center agent could talk to the customer for many minutes, answer all the questions and not break a sweat. It's not about getting rid of call center jobs, it's about making them more productive.

How do you ensure that these virtual agents are working as they should, i.e. not hallucinating?

We use the best and latest LLMs and have provided them with appropriate guard rails. And there are other LLMs that then check them. But a key decision is what you train your model on. So the fine-tuning is always carried out with the customer's data. There is also a human in the loop who can take over the call if necessary.

But of course there are different types of risks with such technology. For example, it's a low risk to have an LLM help you with a marketing text, whereas it's a higher risk to let an AI assistant loose on customers directly. We are still at an early stage in this respect too.

In Mexico, for example, we have an AI project in which a large online university is helping students navigate through its offerings. Many of these questions are easy to answer, such as which courses are on offer or how health insurance works. But of course we have to make sure that everything is right.

The author traveled to the Signal London in-house exhibition at Twilio's invitation.

(mho)