HomeVideos

The state of telecoms & AI in 2026 from Cavell Summit Europe

Now Playing

The state of telecoms & AI in 2026 from Cavell Summit Europe

Transcript

572 segments

0:00

Last week, I sat in a room with some of

0:02

the biggest telecoms providers,

0:03

infrastructure providers, carriers, and

0:06

product builders in Europe. It was a

0:08

one-day conference in central London.

0:10

Every major player was there and five

0:13

things came out of that day that I think

0:15

every founder and every business owner

0:17

needs to hear. Here's what I learned at

0:19

the Cavell Summit Europe this year. The

0:22

Cavel Summit is a conference that brings

0:24

together the telecoms industry across

0:26

the UK and Europe. Founders, CEOs, and

0:29

market analysts all sat in a room for a

0:31

full day of talks, panels, and

0:33

discussions. I founded CircleCloud,

0:35

which is a telecoms company that

0:37

installs and maintains phone systems to

0:39

small and medium-sized businesses in the

0:41

UK. And now I'm working on Wii U, which

0:44

is a modern business phone system, which

0:46

we deliver through Circle Cloud for

0:48

those customers looking for more of a

0:49

managed service offering. and through

0:51

the Wii U website for those customers

0:53

looking for a more self-service

0:54

solution. So, these are my people. These

0:56

are the people that I've been working

0:58

alongside for over a decade. But what

1:00

made this conference different is that

1:02

the conversation wasn't just about

1:04

telecoms. It was about the future of how

1:06

businesses communicate. And that affects

1:08

everyone, not just people in my

1:10

industry. Let me walk you through the

1:11

five things that stuck with me.

1:15

The first one is data sovereignty. And I

1:17

know it sounds dry, but stay with me

1:19

because this one matters more than you

1:21

think. Here's the concept. When you make

1:22

a phone call, a normal business phone

1:24

call, let's say you're in London, where

1:26

does that call actually go? If the

1:28

company providing you with the phone

1:30

service is running their infrastructure

1:31

on a public cloud, let's say they are

1:33

running on AWS and they happen to have a

1:35

point of presence in the US, that call

1:37

will go from your office in London to

1:39

their data centers. It might be in AWS

1:42

in the US somewhere or it might be in

1:45

Europe in which case your data for that

1:47

phone call traverses the UK and it goes

1:50

over the to the US or to Europe or

1:52

another country. The point of this is

1:54

that your data doesn't stay in the UK.

1:56

And for phone calls, you could say this

1:58

is okay. Phone calls are pretty secure

2:00

these days. So you've got SIP, you've

2:02

got RTP, you've got most of the SIP

2:04

calls are running via TLS which is

2:06

secure, but you do have the instances

2:08

where they're not as secure or they

2:10

could be eaves drops. And the point of

2:11

this is that the calls which go to a

2:14

different country, you don't control the

2:16

data policies and regulations in those

2:18

countries. And when you're involving

2:19

phone calls, it's not that much of a big

2:21

issue. But when you're talking about AI,

2:23

then it becomes an issue. Let's say for

2:25

example, envision this. The phone call

2:27

happens, it ends, it's all secure. The

2:29

call recording then gets sent for

2:31

transcription to a service like OpenAI.

2:33

OpenAI is based in the US. The servers

2:35

are based in the US. So that call gets

2:37

transcribed. It's in the US and then it

2:39

gets sent back to you. That journey,

2:41

that call can be intercepted. The US

2:43

could have a subpoena uh sorry, open AI

2:45

could have a subpoena from the US uh um

2:47

Department of Justice and then that call

2:49

is now in the public domain. For

2:50

example, this is a widespread issue that

2:52

is becoming more regulated as time goes

2:54

by. There's a new law that's coming out

2:56

uh over the next few months that relates

2:59

or a new regulation that relates to how

3:02

data is handled in telecoms and in in

3:05

big companies and it's it's one to watch

3:07

out for over the next few months. We've

3:10

put a ton of attention to detail into

3:12

this. We've made sure from the beginning

3:14

that our infrastructure is all hosted on

3:16

our own servers. We run data centers in

3:19

London, data centers in the south coast

3:21

and we even have our own AI servers in

3:24

our data centers. So all the call

3:26

recording transcription, all the

3:27

analysis, all the summarization is

3:30

happening on our own infrastructure

3:32

which we don't need the most powerful

3:34

models in the world to do transcription

3:36

and summarization and recapping phone

3:38

calls. This is a a misnomer. You don't

3:41

need these really powerful models. So

3:42

we've managed to do this on our own

3:43

infrastructure and ma making sure that

3:46

the data stays within the UK making sure

3:49

without even realizing that we are

3:52

sovereign in our data in our data

3:54

handling policies and customers are

3:56

asking about this they're asking where

3:57

does my data go during this phone call

3:59

and for in our case we can say the data

4:01

stays within the UK but maybe your

4:03

provider can't so it's worth asking that

4:05

question the big carriers are setting up

4:07

data centers in the UK they're moving

4:09

their stuff from the public cloud back

4:10

into private cloud and setting up points

4:12

of presence in the countries that

4:14

they're in. Google, they said in the

4:16

conference, is actually doing a deal, a

4:18

partnership with Mistral. Mistral is a

4:21

French AI company that is uh doing a

4:24

great work and because they're based in

4:26

the EU, Google's doing a deal with them

4:28

where they're able to utilize the or do

4:30

a partnership model where they're going

4:32

to be using Mistral's AI infrastructure

4:34

to deliver the Google Gemini product. So

4:36

if mega corporations like Google is

4:38

doing partnerships with a French company

4:40

just to get into the European market in

4:42

terms of AI, there's something

4:44

definitely coming here. So if you're

4:45

running any kind of voice service or

4:47

you're using one, ask the question,

4:48

where does your data go? After all, it's

4:50

your business. It's your data. So you

4:52

should know.

4:55

The second thing that hit me is how AI

4:57

is being discussed. It was being

4:58

discussed at the conference like a

5:00

feature. How's AI going to disrupt

5:02

businesses? How is it going to disrupt

5:03

telecoms? And I'm sat there thinking,

5:05

are we still talking about AI like it's

5:07

coming? It's not coming. It's already

5:09

here and it's already changed how most

5:11

of us work, if not all of us already.

5:12

The way they were talking in the

5:13

conference, I mean, it reminded me of

5:15

like imagine being in the 1900s. I can

5:17

imagine there were companies having

5:19

board meetings and going to conferences

5:21

talking about how electricity is going

5:24

to impact their business. How should we

5:25

navigate electricity? What's our

5:27

strategy for electricity?

5:29

>> Gentlemen, electricity is a force we

5:31

cannot ignore. But the costs, the

5:33

infrastructure, it's madness. The risks

5:34

are too great. We need more data.

5:36

>> It sounds silly because electricity

5:38

isn't a feature. It's everything. It's

5:40

the foundation that everything else runs

5:43

on. That's where I think AI is right

5:44

now. It's not a feature that you just

5:46

turn on. It's not a module. It's the new

5:48

baseline. But the way some of these

5:50

telecoms companies were talking about AI

5:52

in the conference tells you something

5:54

about the industry. I think there's two

5:55

types of telecoms companies, telecoms or

5:58

software companies right now. The first

5:59

type is modern. They live in Kubernetes.

6:02

They've got CI/CD workflows everywhere.

6:04

They've got modern thinking, forward

6:05

thinking developers that are using AI

6:07

every day to code them, build their

6:08

applications. They're deploying it

6:10

either in the public cloud or they're

6:11

deploying it on their own uh

6:12

infrastructure. But in in modern terms

6:15

and modern infrastructure, they move

6:17

really fast. They care about design.

6:19

They care about the user experience.

6:21

They ship fast and they iterate

6:23

constantly. The second type, and I say

6:25

this respectfully, is the traditional

6:27

telco or software company. the ones who

6:29

are running older architectures, the

6:31

ones that were probably developing

6:32

innet, no harm or um you know no insult

6:36

here, but the old type of software

6:39

companies that uh are still slow to

6:41

maneuver. The ones that haven't really

6:42

adopted AI, they don't really trust it.

6:44

They're running an old infrastructure.

6:46

They don't really care about the design

6:48

of the product. They don't care about

6:49

the user experience. They just want

6:51

software that works and they believe

6:53

software to be just a a means to an end

6:55

and they haven't got any attention to

6:56

detail and care about the craft or the

6:58

design behind it. And for their telecoms

7:00

infrastructure, they're running huge

7:01

clusters with archaic technology and

7:04

they're so big and slow to maneuver that

7:06

they haven't really updated a lot of the

7:08

technology where if they did it would

7:10

make a massive impact in their

7:12

operations and efficiency. And look,

7:13

I've got respect for the people that

7:14

have built these big old companies

7:16

because they shaped the industry to

7:18

where it is today. But the world has

7:19

moved on. If you look at companies like

7:21

11 Labs or other startups like that,

7:23

they live in the new era, the new world.

7:26

They don't have technical debt. Well,

7:27

I'm sure they do have some technical

7:29

debt, but nothing like the old

7:30

corporates and they move at completely

7:32

different speeds. Traditional Telos,

7:34

they tend to operate more like the

7:35

second type. You know, they're slower.

7:37

They're big, massive corporates with

7:40

legacy infrastructure. Their appetite

7:42

and capability for rapid change is

7:44

lower. And that creates a gap. The

7:46

companies that understand that AI isn't

7:49

a feature, that it's the new

7:50

electricity, they're the ones that are

7:52

going to win, not because they're

7:54

smarter, but because they're faster, and

7:56

speed in this market is everything.

8:01

Now, the third point, and this was by

8:03

far the biggest point of conversation at

8:06

the event, and they actually put at the

8:08

beginning of the conference, they put on

8:09

the screen a question for everybody to

8:11

answer. and you had the ability to scan

8:13

a QR code and you grab your smartphone

8:15

and then you could answer the question

8:17

on your smartphone. So everybody or most

8:19

people at the conference answered the

8:20

question and the question was what do

8:22

you want to get out of this conference

8:23

today? What was the what's the topic

8:25

that you wish was discussed or you want

8:27

to be discussed at this conference and

8:29

the most without doubt the most common

8:32

answer was voice AI. How is voice AI

8:35

going to impact my business? How is how

8:37

do I adapt and produce a voice AI

8:39

solution that works for customers and

8:41

and and provides a solution? Who are the

8:43

best voice providers? Voice AI

8:45

providers. It was all about voice AI.

8:47

The organizers of the conference

8:48

actually planned the conference mostly

8:49

around sovereignty, data sovereignty and

8:51

other related pieces. AI was there, but

8:53

voice AI may be hidden by surprise or

8:55

something, but voice AI was definitely

8:57

the topic of the day. And it makes sense

8:59

because voice AI is already disrupting

9:01

how businesses communicate with their

9:02

customers. Let me give you two examples

9:04

that we shared at the conference. The

9:05

first one comes from a company called

9:07

Bamwith. It's a US carrier which has a

9:09

strong presence in Europe and they've

9:11

built a voice AI agent that is capable

9:14

of performing action similar to what we

9:16

want to do with Ringup built into Wii U.

9:18

They've built it at a higher um business

9:21

a bigger business level more of a

9:22

corporate level and they installed it.

9:24

They gave two examples one of them was

9:26

at a hotel chain big hotel chain in the

9:28

US which had a problem with their front

9:30

of house um employees. that people that

9:33

were working in front of house weren't

9:35

able to answer the phone in busy

9:37

periods. And what was happening is that

9:39

guest guests were calling down to guest

9:41

services maybe to place an order for

9:44

room service or to book a table for the

9:47

restaurant that evening or whatever it

9:48

may be and the calls were just not

9:50

getting answered and the impact on guest

9:52

experience was devastated for their

9:54

devastating for their chain. So what

9:56

they did is they implemented voice AI

9:58

before it reached the human. And that

10:00

voice AI had access to the booking

10:02

system. It had access to the system, the

10:04

guest services system. It could place

10:06

orders for room service. It could place

10:08

um change reservations and and check

10:10

reservations etc. And what it did is it

10:13

allowed the humans within the reception

10:16

to attend to other guests and it

10:19

drastically improved the guest

10:21

experience within their hotel chains.

10:23

The second example they gave was a busy

10:26

chain, a chain of busy restaurants that

10:28

at peak hours didn't have the waiters

10:30

available to answer the phone calls from

10:32

other restaurant guests calling in to

10:35

either book a table or to change a

10:37

reservation or to inquire about the

10:39

menu. And in fact, I think to myself,

10:40

who these days would call a restaurant

10:43

rather than just going onto the website

10:45

or going on to Google and then booking

10:47

the table from there? And in fact, I am

10:49

one that would do that if I'm driving or

10:52

if I'm busy doing something else. I just

10:53

grab my phone and just dial the

10:54

restaurant quickly or I tell Siri to

10:56

dial the restaurant and I try and book a

10:58

table there if I'm doing other things. A

10:59

lot of the times when I do that I don't

11:01

get an answer and that's because they're

11:03

busy attending other customers. So what

11:04

voice AI can do once you've connected it

11:06

through to the booking system is it

11:08

would allow the the caller to inquire

11:10

about the menu. It would inquire about

11:12

uh what tables are available. It would

11:14

arrange the booking all done with an

11:16

automated agent without needing to speak

11:18

to a human which means that the humans

11:20

can then attend other customers within

11:23

the restaurant improving the overall

11:25

customer experience. Now here's the

11:27

interesting tension that I detected at

11:28

the conference. They were talking about

11:30

these new startups, these new tech

11:33

companies that are creating all these

11:34

voice AI products that are competing

11:37

with the Telos because what they're

11:38

doing, they're putting the voice AI

11:39

product in front of the tele telco

11:42

solution, in front of the unified

11:44

communications platform or the phone

11:46

system and extracting margin from the

11:48

telco providers. And I think that

11:50

creates a differentiation between the

11:51

two types of companies. You've got these

11:53

telco big telco operators that are slow

11:55

to move and they provide the telecom

11:57

services but they don't really provide

11:59

those value added services on top and

12:01

then you got the brand new startups who

12:02

don't know anything about telecoms. They

12:05

provide the solution for the customer

12:07

and that solution you know might be

12:09

communicating through uh an API to

12:12

something like Twilio so they can make

12:13

and receive phone calls. They don't know

12:15

anything about telecoms but they know

12:16

about the customer and they end up

12:18

winning the the deal. And how much of a

12:21

better solution would they be able to to

12:23

run and provide if they were aware or at

12:26

least if they knew about telecoms as a

12:28

whole? If they knew about session border

12:30

controllers, if they knew about how uh

12:32

soft switch works and how to routt calls

12:34

and how to run phone calls on premise,

12:36

they would provide ultimately a much

12:38

better level of service. The the call

12:40

quality will be much better. The

12:42

troubleshooting will be much faster. the

12:44

ability to run many more phone calls

12:46

will be better and more stable overall

12:48

long term at scale. And finally, it will

12:50

probably be cheaper for them overall,

12:52

but they need the skill set for this.

12:54

And we're positioned in a unique place

12:56

because we have the software skills. We

13:00

have the quick pace of action. We move

13:02

quickly and we also have the telco

13:04

experience. We have the soft switch

13:06

experience. We are running our own

13:07

network and we're positioned within that

13:09

intersection of a modern software

13:11

company and a traditional telco right in

13:14

the middle. That middle ground that gap

13:16

is where we are positioned and that's

13:17

where we think the opportunity is.

13:20

The fourth point, and this was a clear

13:22

one delivered at the conference, is that

13:23

people believe that the voice services

13:26

as a traditional method, you know, when

13:27

you you have a 10 to 15 pounds per user

13:30

per month license that comes bundled in

13:32

with a,000 minutes or something, that

13:35

service, the price of that service is

13:37

trending towards zero. Eventually, maybe

13:39

it will even reach zero. So, how do

13:41

telecoms companies make money? How can

13:43

they make money when that traditional

13:46

service is going to be obsolete? And

13:48

it's by value added services. What are

13:51

value added services? It's the

13:52

integrations. It's the overall customer

13:55

solution because it's one thing

13:56

providing a phone system voice license

13:58

where they pay per user a month just to

14:00

be able to make and receive phone calls.

14:01

And it's something else providing that

14:04

plus on top data analytics so you can

14:06

plug in your CRM or whatever system you

14:09

use your financial system into the

14:11

platform so that you can present data in

14:13

the form of dashboards and reports not

14:16

just of your phone system but also of

14:18

your business uh applications in one

14:20

single dashboard or scheduled report.

14:22

It's something else to be able to have

14:24

additional services and tools such as,

14:25

for example, Ringup giving you the

14:27

facility to implement voice AI agents

14:30

within your phone system to root calls.

14:32

Not just agents that can answer phone

14:34

calls and take voicemails, but agents

14:36

that can actually do and perform actions

14:38

by plugging into the current CRM that

14:40

you integrate with because of the

14:41

integration that we've built through EU.

14:43

And the third is integrations

14:45

themselves. because all businesses or

14:47

most businesses use a CRM or a financial

14:49

system or some kind of platform to

14:51

record and save and manage their

14:54

business. So, integrating into these

14:56

platforms allows a productivity boost,

14:59

an efficiency boost, and allows

15:01

customers to do the best work possible

15:03

by having access to the best technology

15:06

for their business. So, why is it that

15:08

the cost of voice licenses is going to

15:09

continuously trend down to zero? I think

15:11

it's down to services like WhatsApp, for

15:13

example. WhatsApp calling, it's end to

15:15

end. It's free. People are used to just

15:17

picking up their phone, going into

15:18

WhatsApp, and calling a contact through

15:20

WhatsApp, either with audio or with

15:21

video. And there's no cost to that call.

15:23

But why is there no cost to that call?

15:25

It's because it doesn't touch the public

15:27

telephone network. You can't just pick

15:28

up and dial a number on there. And when

15:31

it comes to businesses and when it comes

15:32

to communicating with businesses,

15:34

businesses still have phone numbers. And

15:36

that's still the universal method to

15:38

communicate with businesses through

15:39

voice. So at Wu, we've realized that the

15:41

commodity is the phone line. The value

15:44

is the products you build around it and

15:46

the solutions that you solve for the

15:48

customer.

15:51

And the fifth thing, this is less about

15:53

what I learned at the conference and

15:54

more about what the conference reminded

15:56

me of. We're in the process of taking

15:57

Wii U and delivering it as a SAS as a

16:00

software as a service solution. And we

16:02

need to find a way to differentiate the

16:03

product. We need to find a way to

16:05

specialize because in a crowded market

16:07

and telecoms is a very crowded market.

16:09

If you don't differentiate, you're just

16:11

invisible. I remember reading a brochure

16:13

around 10 years ago and there was a line

16:15

from a a VOIPE company that was selling

16:17

cloud communications and they said how

16:20

to stand out from the cloud and I

16:22

thought that was quite clever but it's

16:23

even more relevant now. So how do you

16:26

stand out in a crowded market? The way I

16:28

think about it is this in order to stand

16:29

out in a crowded market you need to

16:31

understand the customer and understand

16:33

their workflow. And in order to

16:35

understand their workflow, you need to

16:36

really know how they operate day-to-day

16:39

and know what problems they have so well

16:42

that you can create a solution that

16:44

removes friction from their problems so

16:46

that they can solve them faster and

16:49

better with your solution. And the only

16:51

way you can do this is one by

16:54

verticalizing meaning that you

16:56

specialize in a particular industry and

16:58

you solve a problem that that industry

16:59

has that because you understand the

17:01

industry you're able to solve that

17:02

problem really well. or two, you

17:04

specialize in a particular solution that

17:06

might apply to multiple industries, but

17:09

you solve the problem within that

17:10

solution extremely well. And with W, you

17:12

see, we're currently evaluating both of

17:14

those options right now. But to us, the

17:15

principle is clear. In a noisy market,

17:17

the businesses that will succeed are the

17:19

ones that are able to create a solution

17:21

that is super laser focused on solving

17:23

one problem better than anybody else in

17:25

the world is. The businesses that are

17:27

known for something specific, not

17:29

something generic, not we do everything

17:31

for everyone. No, we do something really

17:33

specific for one specific type of

17:35

profile, but we're the best in the world

17:36

at it. And that's what we're going to

17:37

focus on in Wuci. Create that unique

17:39

proposition that is solving that one

17:42

problem extremely well for a particular

17:44

segment. And once we've made that

17:46

decision of what segment that's going to

17:47

be, we're going to focus all our

17:49

marketing efforts around that.

17:53

So, five things from one day at the

17:55

Cavell Summit Europe. Data sovereignty,

17:57

it's real and it's actually becoming

17:59

regulation really soon. AI isn't a

18:01

feature. It's actually the new

18:03

electricity. Voice AI is the biggest

18:05

opportunity in telecoms right now.

18:07

Traditional voice pricing is probably

18:09

going to zero. And the only way to win

18:11

in a crowded market is to specialize.

18:13

None of these exist in isolation. They

18:16

overlap. The company that builds a

18:18

communication platform that has all

18:19

these at its core. It has AI at the

18:22

core. who are able to solve really

18:24

specific problems better than anybody

18:26

else in the world can who building their

18:28

systems and infrastructure in a

18:29

sovereign way. So it handles data in a

18:32

sovereign way and who specialize in a

18:34

particular vertical or in a particular

18:36

solution. Those are the companies that

18:38

will win and that's what we're building

18:39

at WC and that's I find these

18:41

conferences super motivational and super

18:44

interesting. They remind me that we're

18:45

heading in the right direction. If you

18:47

want to follow along with what we're

18:48

building, I share the lessons of what

18:50

I'm doing and what I've learned and what

18:52

we're building. So, subscribe. And if

18:53

you want my founders operating system

18:55

framework that I've built, click the

18:57

link in the description. This was my

18:59

overview of what I learned at the Cavell

19:01

Summit Europe last week. See you in the

19:03

next one.

Interactive Summary

Loading summary...