CXTech Week 31 2023 News and Analysis

The purpose of this CXTech Week 31 2023 newsletter is to highlight, with commentary, some of the news stories in CXTech this week. What is CXTech?  The C stands for Connectivity, Communications, Collaboration, Conversation, Customer; X for Experience because that’s what matters; and Tech because the focus is enablers.

You can sign up here to receive the CXTech News and Analysis by email or by my Substack. Please forward this on if you think someone should join the list. And please let me know any CXTech news I should include.

Covered this week:

  • Rethinking LLMs with Rob Pickering
  • TADSummit 2023
  • Giovanni (Johnny) Tarone gives a taster for what is coming in his A2P Masterclass at TADSummit
  • TADHack Global, 21-22 October 2023, dev resources coming soon
  • Developers do not want Network APIs, Just keep out of their way, NNIs are another matter
  • The latest 5G innovation: Even more price increases
  • Route Mobile admits failed 6-year bid to establish strong presence in booming US CPaaS market
  • RTC Security Newsletter July 2023
  • The Great SPAC Scam
  • Becoming a Platform is Not a Strategy: Geoffrey Moore
  • People, Gossip, and Frivolous Stuff

Rethinking LLMs with Rob Pickering

I think I’m now willing to ignore my inner sceptical technologist and predict that lots of people are going to be using LLMs to automate away the cost of many voice communications.

Rob Pickering
A younger AI version of Rob not listening to his inner sceptical technologist 😉
Source: mid journey with prompt “<picture of Rob> with a small angel on the left side and a small devil sat on the right side, cinematic, global illumination.” Couldn’t get an angel or devil on left and right or on his shoulders, I thought this one was good enough.

Old position:

My distilled position on the technology can be characterised as: “It is supercharged pattern matching and pattern generation, nothing more”. This is derived from my understanding of the simplicity (and inherent difficulty in scaling) the 30 year-old computer science techniques used to build models.

It lead me to write this article earlier in the year… What can GPT do for your business right now? Don’t read it now. It isn’t outright wrong, but it does focus in a very limited way on content generation. I missed by a mile the transformative nature of the instruction following and summarisation capabilities of recent LLMs for conversational dialog.

New position:

I think I’m now willing to ignore my inner sceptical technologist and predict that lots of people are going to be using LLMs as a tool to automate away the cost of many routine communications that we currently use humans for. It isn’t just copywriters who will face threats and opportunities. Doughnut sellers are going to need to embrace it too!

The societal implications of this are going to be immense. So is the economic and human value which means we aren’t, even now, going to get the genie back in the bottle. I think it is going to be incumbent on those of us harnessing the technology to focus on the true value proposition to society whilst guarding against negative externalities.

Rob has a LLM model you can play with, https://llm.aplisay.com/landing.

He’s going to make this tool part of the resources available at TADHack through sponsor Jambonz. I’m hoping by next week we’ll have the Jambonz developer resources available.

And, Rob’s going to be presenting at TADSummit:

LLMs on the telephone: useful tool, or hallucinating danger to humanity?

Rob Pickering, Software: Real Time Communications and Machine Learning

  • Rob talks about Aplisay’s experience developing an open source LLM voice tool and the journey that has taken them on in evaluating practical use of the technology.
  • He will describe how it was built and then give his perspective on questions it is already starting to answer like:
    • What are the advantages of the LLM approach over existing dedicated conversational interface AI stacks?
    • Where can it go wrong, what are the dangers?
    • Do some applications initially make more sense than others?
    • What techniques could we use to harness the highly capable summarisation and instruction following of LLMs whilst containing their tendency to go spectacularly off-piste?

TADSummit 2023

It’s time to start planning your trip to Paris on the 19/20 October. We have an impressive no-BS agenda coming together. If you follow me on Linkedin you’ll see there’s going to be lots of juicy conversations at TADSummit.

We’ll be discussing the decline of 2FA; passkeys, FIDO2, and WebAuthn will be drivers for the decline. We have an excellent presentation from Dan Jenkins on that topic. There is simply no event in telecoms that brings people working at the coalface together to have frank, insightful, and deeply understood discussions – no waffly marketing-level stuff.

Thank you to our sponsors Unifonic, SIPhub, TelecomsXChangeStacuitySTROLID, and jambonz for making TADSummit possible.

Passkeys, FIDO2, WebAuthn… What does it all mean?

Dan Jenkins, Founder at Nimble Ape Ltd, Director at CommCon Events Ltd, Director at Everycast Labs Ltd

  • Protecting your online identities has become a huge pain in the rear.
  • 2FA in the forms of SMS, TOTP, Password Managers, FIDO, U2F and now Passkeys – it’s just too much.
  • Unless you’re a ‘geek’ how can you hope to protect your online identities in the best way possible?
  • Passkeys are here to help answer that question. This session takes a look at Passkeys, how they work and whether you should all throw away your security keys in favour of them.

Giovanni (Johnny) Tarone gives a taster for what is coming in his A2P Masterclass at TADSummit

Oh wow! When Giovanni said he was giving a taster for what he was going to cover in his A2P Masterclass at TADSummit, I didn’t realize it was going to be this! I am flabbergasted that there’s so much intrigue for such an apparently straight-laced German company, tyntec.

Giovanni has so many stories to share, it’s important learning for everyone in the industry. There are too many skeletons hidden in closets, and in the limit its the end customers (us) that pay, whether it’s through SPAM, robocalling, or inflation from all these A2P SMS price rises and nebulous processes for a decades’ old service.

The US A2P market is not as open as we may think. Check out Rajdip Gupta of Route Mobile Limited‘s CNBC article where he admits, “their failed 6-year bid to establish strong presence in booming US CPaaS market” as discussed in this newsletter. Route Mobile’s last quarter was $122M, they are an impressive company with a low cost base, and they could not profitably break in to the US market! The failure of the court case Giovanni refers to around the Iris Wireless / Tyntec / Syniverse peering agreement is directly linked to Rajdip’s struggles, https://alanquayle.com/2017/02/tyntec-an-interesting-development-in-the-us-telecom-api-market/. The unusual investment from Twilio in Syniverse and Syniverse’s recent downsizing, are also linked. There’s something funny going on under the US A2P sheets.

Looking beyond the US, now onto the Fantasy ‘CPaaS’ M&A League, we’re going to have fun with this at TADSummit. Customer concentration was always a concern for potential buyers of tyntec. Would a roll-up with MessageBird give greater scale across Europe? I’m not sure given all the cash raised by MB, and current investor frustration. Perhaps a roll-up with Soprano Design to create a truly global programmable communications company with a solid high margin? I see the technology / network of GMS – Global Message Services and tyntec as similar, and they are to an extent complementary geographically, though GMS is becoming a monster, growing faster than anyone in this space. That could make a new independent SMS aggregation powerhouse as a roll-up.

Obviously, TwilioSinchInfobipSyniverse (with Twilio’s approval given their board seats in Syniverse) would all love the network / technology assets as an addition to their’s. And what about Mitto? Bigger than tyntec, on Linkedin I see significant people churn between them and tyntec. There are close to one hundred more companies I can add, through they tend to be a little more focused, e.g. Unifonic, Cequens, Vibes, Macrokiosk, Clickatell, CM, Plivo, Gupshup, etc.

TADHack Global, 21-22 October 2023, dev resources coming soon

TADHack (tadhack.com) is the largest and longest running hackathon focused on programmable communications / telecoms. Since its founding in 2014, TADHack has always been hybrid, you can take part in-person or online.

The purpose of this note is to remind you to save the dates (21-22 Oct, please add them to your calendar) and let you know the developer resources are coming soon. With Stacuity we’ll have SIMs to distribute, we’ll ask you to check out the dev resources sooner this year to ensure the SIMs get to people who need them before the hackathon starts.

At TADHack you’ll learn about important new technologies, share your hack and skills with the world, and potentially create a solution to an important problem that could become your side-hustle or even main job, like many other TADHackers. This is our 10th year of TADHack, we guarantee you’re going to have fun!

To win cash prizes you must hack on the global sponsors’ technologies. This year we have:
STROLID. vCon is the new global standard, a ‘PDF for conversations’;
Stacuity. Mobile connectivity for IoT;
Jambonz. Open source voice platform; and
Radisys. An amazing drag and drop development platform, for loads of resources including communications, speech recognition, natural language processing, and video analysis.

Go to TADHack.com and register!

Developers do not want Network APIs, Just keep out of their way, NNIs are another matter

Mike Dano had a piece on “T-Mobile this week announced that software developers can use network slicing on its standalone (SA) 5G network to build video calling applications with more consistent uplink and downlink speeds and lower latency. The result, the company said, could be more reliable video calls that may not freeze or skip. T-Mobile noted that companies like Dialpad Ai, Google, Webex by Cisco, Zoom and others have signed up to test out its new capability.”

A network API for a rare problem where congestion in the 5G RAN or network is likely not even in the top 5 problems that could cause the video glitch shows an industry trying to desperately trying to find problems for its solution.

The more likely causes are the device / OS and all the apps running in the OS, the conferencing client and its battle for a slice of the device’s resources, the other video conference participants and all their multiple paths and resources to the conference service instance, the conference instance and its contention for virtualized resources, etc.. To think a broadband link of hundreds of Mbit/s could be the principal cause of the video glitch is a little delusional.

Network QoS is not new, PCMM (PacketCable Multimedia) offered dynamic network QoS with APIs about 20 years ago on fixed cable broadband. There’s much mobile operators could learn from the cablecos’ experiences. I helped Camiant, a leader in PCMM, break into mobile operators during the policy control and charging hype cycle.

Network as a Service (NaaS) is a mature existing market, it was covered at TADSummit in 2021 by Liang Dong of Epsilon (now KT) https://blog.tadsummit.com/2021/05/24/delivering-the-future-of-networking-with-hyper-scalable-connectivity-liang-dong-epsilon/. NaaS can be extended to Mobile broadband, but that’s more a network:network interface rather than API. Customers and developers do not care about fixed or mobile networks, they just care about internet access that works reliably and transparently. Keep the Network API out of their way!

Note the programmable telecoms industry is focused on service APIs: telephony, messaging, etc. Cable cos learned a network API just gets in the way for customers and developers. Focus on the NNI not the API when it comes to QoS, NaaS is a mature business.

The latest 5G innovation: Even more price increases

An excellent article from Mike Dano.

In the early days of 5G, proponents promised the technology would unlock a wide range of new revenue streams for innovative operators to plunder. And in the intervening years, companies like Verizon and AT&T touted the 5G potential around services like network slicing and edge computing.

But today, as T-Mobile, Verizon and AT&T put the finishing touches on their initial 5G network buildouts, they’re turning to a much more straightforward 5G monetization strategy: price and fee increases.

As noted by 9to5Mac, both Verizon and AT&T are now increasing the cost of their legacy smartphone pricing plans by around $3 per line per month. “This increase will allow us to continue to deliver the great wireless service you expect,” according to AT&T.

Those are just the latest price and fee hikes enacted by the nation’s big 5G providers. For example, Verizon is increasing the price on its 5G home broadband service to new customers by $10 per month. Separately, T-Mobile, AT&T and Verizon are reducing their autopay discounts for customers who use credit cards. Even some prepaid providers, like T-Mobile’s Metro, are getting into the act.

Route Mobile admits failed 6-year bid to establish strong presence in booming US CPaaS market

This is an excellent and most importantly honest review of the challenges in entering the A2P SMS market from Rajdip Gupta of Route Mobile Limited. Proximus recently bought Route Mobile, covered in CXTech Week 29 2023.

The North America A2P messaging market is tough and getting tougher, looking at you The Campaign Registry. I’ve discussed the strange investment between Twilio and Syniverse and what may be going on under the covers. The tough situation Syniverse finds itself with the carriers pulling the strings, it should be printing money given its unique and critical position, not downsizing.

Telesign does not provide direct access to the US carriers, rather an amazing customer list across most of the web-based service providers that Route Mobile can sell its broader portfolio to.

Looking to Europe I see a more open environment where access to the carrier hubs is straight forward, yes, it can be cheaper to go direct for some of Telefonica’s LATAM properties. And if your traffic misbehaves you get cut off, the carriers do KYC (Know Your Customer) in Europe.

At TADSummit in Oct 19-20 you’re going to experience more of the same honesty and openness demonstrated by Rajdip in this article.

RTC Security Newsletter July 2023

It’s the newsletter you can not afford to miss, a few highlights,

Congratulations to Enable Security for being invited to present at Bloomberg RTC Summit about VoIP and WebRTC attack surface!

Also congratulation to Gavin Henry who released SentryPeer which helps prevent VoIP attacks and toll fraud. It does this by providing APIs that allow users to query for phone numbers or IP addresses. Specifically, the APIs are able to tell if a phone number is considered fraudulent or if a source IP is a known attacker address. It does this by relying on the SentryPeer honeypots that are crowd-sourced and feed in this data.

The service has a business model and is also available for free for those that contribute data through their own SentryPeer honeypots. It is open source and quite an interesting initiative.

Similar note-worthy efforts that come to mind and are also part of the RTC security community are:

In related news there was a huge spike in REGISTER traffic attacking SIP servers, many coming from the Ingate Separator SBC:

  • Fred Posner posted on Linkedin about traffic involving open SIP proxy abuse, that is being seen on the honeypots
  • APIBAN naturally blocks the IPs for the SBCs that are being abused and similar traffic
  • This points back to Ivan’s blog post which we covered last month about the topic

Vishing with “Letscall” using VoIP and WebRTC

Threat Fabric released a report about Android malware targeting individuals from South Korea. What is interesting is that it acts as a voice traffic router by redirecting incoming and outgoing calls. Depending on how the malware is configured, it might redirect the calls to a call center controlled by the criminals. To do this, the mobile application makes use of VoIP and WebRTC and abuses the legitimate service ZEGOCLOUD.

The Great SPAC Scam

If you’ve read this weblog for a while, you’ll know I have a relatively negative view of SPACs, Special Purpose Acquisition Vehicle. This article is excellent, and includes an Excel model to play with, it breaks down the mechanics of a SPAC. I’ve seeing many more stockholder class action lawsuits as a result of SPACs fleecing regular shareholders.

SPACs are yet another alternative to the traditional IPO process and can be faster than IPOs. Unlike IPOs, however, the Sponsor gets a 20% stake, called a “Promote,” and there’s much less regulatory scrutiny.

As a result, the Sponsor could walk away with millions or tens of millions of dollars even if the SPAC performs horribly and the share price plummets, while normal investors will lose everything, and that seems to happen a little too often in my opinion.

Becoming a Platform is Not a Strategy: Geoffrey Moore

This is an excellent 6 step guide.

  1. Start with a killer app. OK, it is a big ask, but the first qualifying step for becoming a platform is to secure ubiquity first. No one is going to buy into a platform that they have to help make ubiquitous. Instead, you need to bring to market an offering that gets very widespread adoption as a point product and then leverage that position to kick off the next step.
  2. Assemble a portfolio, evolving into a suite. This capitalizes on your ubiquity by capturing adjacent applications and use cases. One or more of these products will likely come from acquisitions, and that is what will trigger your first explicit investment in platforming.
  3. Platform for internal productivity. You migrate to a platform architecture with APIs initially in order to simplify the maintenance of your suite as well as to unify the user interface into a potpourri of products. You don’t charge for this, funding it instead from the savings you get internally.
  4. Open up your internal platform to major customers. These are enterprises that have the in-house resources to build differentiating, proprietary applications on top of your platform. They further pressure test your APIs, as the things they are looking to do are explicitly not on your roadmap. Here you do want to charge a platform licensing fee in order to fund the migration of the platform from an internal productivity tool to an external product. This is when the platform gets a product manager, a product roadmap, and an ROI-based development plan.
  5. Recruit ISVs (Independent Software Vendors) and SIs (Systems Integrators) to your platform. These folks will come if, and only if, you have made a market for their products or services. Second-tier players will lean in to help here as best they can because they see the possibility of garnering a first-mover advantage. Because platform plays are gorilla games only (not much use for whatever platform came in second), first-tier ISVs and SIs will only commit when it has become clear you are the gorilla in the space. You can mitigate this risk with a market segment domination strategy where you are the “gorilla in the niche” if the market segment is sufficiently attractive to warrant their participation.
  6. Support the build-out of the rest of the ecosystem. Once there is material ISV and SI support for your platform, then you can secure your position for the long-term by encouraging secondary systems and services, be that platform administrators, training programs, or support groups. Microsoft did this masterfully in the 1990s for their Windows NT platform, and Salesforce has done the same for its CRM suite.

That’s the playbook in six steps. No step is easy, but none is miraculous. The main thing is to take things step by step and to secure your position at each stage before investing in the next.

People, Gossip, and Frivolous Stuff

Congratulations to Martina Saric who is now Managing Consultant at Business Transformers eG. I’ve known Martina since she was heading up service enablers at Magenta Telekom.

Claire Marshall is now Fractional CMO Services at Onami Consulting. I’ve known Claire for over a decade, since her time with NeDev, which had one of the first WebRTC conference services.

Pankaj Fichadia is now Head of Cloud Governance and FinOps at NAB. I’ve known Pankaj throughout his time at NAB.

Harold Vance is now Director, Customer Success at Microsoft. I’ve know Harold since his time at Tropo.

Frank Fischer is now  Vice President Information Technology at Adalvo. I’ve known Frank for over a decade, since his time at DT’s Developer Garden. I think the organizational learning in many telcos from their developer API experiences has not been retained. Frank should be running their OneAPI 2.0 program, it would be very different if he was.

You can sign up here to receive the CXTech News and Analysis by email or by my Substack.