Everyone is seemingly all-in on AI at the moment. Claude and Claw. Prompts and process. Agents and apps.

I am too (see blog posts passim). But, I’m also all-in on another trend which is transforming the world around us. It is not as well-covered, but is going to change the way in which you do business and what your digital experiences are like as a consumer.

Sovereignty.

In a technological context, “sovereignty” means being highly specific about the source of something. That source might be where your data is hosted, or where a piece of software comes from. Data sovereignty is all about the jurisdiction where data is stored, how it is hosted when it’s there, how the data got there, where the data is going, and all of the in-person and digital support that the data has when it’s there.

This trend has come about through a perfect storm. The first is, of course, geopolitics. The world is never perfectly stable everywhere and there is a general feeling in the west that instability is growing, whether perceived or otherwise. The second is a repatriation of data, particularly from the public sector, where contracts — and increasingly national laws — stipulate that data about citizens, for example, must be hosted in the same country and cannot leave national borders. The third is perhaps a more reactive feeling of “why should we pay company X based somewhere else to do this, when we can do it ourselves” — or at least pay a company that is based in the same jurisdiction or region.

You may be thinking that in our global age — does any of this really matter? Well, yes, it does. Hosting data about citizens outside of their jurisdiction invites all sorts of security issues, notwithstanding the general attack vector for that data being incrementally greater. You might not want your tax history being stored in a server on the other side of the world, and for that data to be subject to the laws of that country, rather than your own. As a government agency or department, you’d want to keep your data in the same country, and to be able to keep an eye on it (as it were) with no ability for it to go anywhere else. Hence, sovereignty.

It’s not just about where the data is stored, either, but who is storing it. France has some of the strictest requirements around this. Called SecNumCloud, the requirements in a nutshell are that public sector data must be stored in France, available nowhere else, and stored by a company whose ultimate owner is also a French company. The sovereignty “spine” runs top to bottom.

One might think that such requirements knock out the hyperscalers but not necessarily. Google has partnered with French company Thales to launch a joint venture called S3NS, where Thales basically runs a standalone iteration of Google Cloud, in French data centres, to architectural specifications set by Google. It ticks the sovereignty box but also allows all of the advantages of cloud hosting to continue. Microsoft has done the same in France and Germany to offer what it calls National Partner Clouds for an in-country Azure offering from sovereign partners, and IBM has launched Sovereign Core which is a comprehensive control plane for sovereign cloud and data management.

Countries where sovereignty requirements are weaker than those of France still have a big ask of cloud providers. Germany’s requirements, called C5, sit alongside those of many other countries — including those outside of the EU, such as Australia — in requiring sovereign hosting to meet a set of standards, published by a nominated government agency. That might seem bureaucratic but it does help sovereign standards to be robust, as that government will have a consistent defintion of “sovereignty” for cloud and software vendors to adhere to.

AI of course is very much part of the sovereignty debate. The provenance of LLMs is hugely important (I wrote a post on it here) in terms of how your data is understood, processed, stored… and, of course, what data the LLM is trained on. When you make AI activities agentic and then supercharge them — see Openclaw — sovereignty becomes a priority, as your data is thrown around all over the place.

All of this challenges the concept of globalised platforms. If you have a multinational business which uses global platforms for such processes as HR or CRM (with AI services baked into such platforms), then you might be using one of the popular cloud-based vendors. The same goes, of course, for AI services.

But, where are they hosting your data, and does your data contain any records which are subject to national requirements? If so, now’s the time to have the conversation with the vendor as to how they can meet them, and if not, what your Plan B is, in order to avoid any jurisdictional breaches. You can only be as “sovereign” as the least sovereign part of your IT estate.

Sovereignty doesn’t just affect large businesses turning over gigabytes of data — it dramatically affects the user experience too. I gave the answer of health data earlier but think of where your personal photos are stored. Do you know? Does it matter whether you know or not? There are plenty of discussions happening at the moment around potential government laws regarding backdoor encryption and VPN blocking, and whatever opinion you may have of these points, they are all about the ownership of, and access to, your data… in other words, sovereignty.

But, as I mentioned earlier, this is not just a data issue. Some public sector organisations across Europe, particularly in Germany, have switched away from popular commercial tools to open source and/or homegrown alternatives. It’s early days to really understand the effect of such changes but the wind of change is certainly blowing for some.

Sovereignty is such a big and important topic that it’s hard to say everything about it in this short article. I have been conducting a lot of research into this topic and find it fascinating. Drop me a line if you’d like to know more; it’s an ever-changing but critical topic that will be part of all of our socio-technological futures.

I was recently in an internal discussion group where a colleague asked how they could download a bunch of PDFs at once from a website. The PDFs were published across a many child pages and subsections, so File -> Save would have been monotonous. Additionally, the web pages containing the PDFs had a file extension specific to a particular web server. This mean that the obvious candidate for downloads — Wget — just couldn’t work.

That problem then turned into an AI challenge. Could an AI code generation service produce a Python script to recursively download PDFs?

The AI service not only produced bug-free code that ran first time but also offered me more options when running it, including the URL to start from and how many levels deep the scraping/downloading should go. I ran the code and handed the PDFs over. Job done.

AI code generators and managers like Replit, Google’s Gemini, and IBM’s Bob can produce lightning-fast code and, in some cases, one-click deploy the code to a cloud provider and save the code to an internal repo or shared drive. It’s this integration that makes such offerings all the more easy and enticing to use.

Showing these offerings to people not aware of AI’s possibilities honestly makes their jaw drop. The aforementioned PDF downloader took about 2 minutes to produce, start-to-finish, inclusive of one point where the code generator corrected itself in order to ensure that the downloader correctly worked.

This led me to ponder.

What these code generation/management offerings are doing is exploding the myth that application development is a technical phenomenon. It is becoming a conversational one. If you can express yourself through the written word, then you can build and deploy an app. You can do it in your lunch break. You’ve probably seen such offerings in places like YouTube pre-roll ads which indeed make such a promise.

And, because it is now a conversational phenomenon, there is no barrier to entry. Of course, it helps if I have at least a smattering of a technical understanding, but it isn’t critical, and such services will only make their user interfaces easier over time, to the extent that no-one will be nervous about making apps (not that they should be anyway).

It creates a massive, and I mean massive opportunity to create true inner source opportunities within an enterprise.

Historically, internal web applications were mostly the preserve of centralised teams as we know. There was an Intranet team, various teams to integrate applications into the Intranet universe, and wider networks of web authors.

We can now completely democratise that network.

Imagine a situation where practically anyone in an enterprise can produce an app with a conversational interface. They can save the code to a network drive and one-click-deploy to an internal application server. The apps could range from managing rota shifts to staff competitions and vacation planners, or whatever else employees may need to quickly get over a problem or blocker.

These are not apps really, they are applets. Of course, there would be a publishing and approval process to moderate code and technical quality and to prevent duplication, an appropriate guardrails would constain what code is output and why, but otherwise… why prevent the blooming of a thousand flowers? Let those closest to the problem develop the most effective ways of resolution.

The internal application server could be architecturally fronted by a conversational search engine to ensure that apps are exposed to teams and the enterprise at large, so through conversation you would know what’s available. The fostering of this inner source culture would immediately fire up new communities of practice across the enterprise and there would be active encouragement and support for them to develop new and better applets.

Enterprises can often only move forward as quickly as their slowest part and the challenge (= opportunity) here is, to foster a culture that allows this to happen, with platforms and technology that are enterprise-grade and security-hardened while also being easy to use. Some of the most innovative stuff happens at the edges. I remember finding out about a colleague at my then-employer who had built an application that lived on a 486 PC under his desk which showed live electricity network outages on a map. We took it, production-ised it, and it won a government award.

AI code generation and management gives practically everyone and anyone an opportunity to develop solutions that meet real problems in the workplace, but as I said above, it’s a cultural challenge as well as a technological one. However, without it, you get shadow IT and shadow AI — uncontrolled, insecure, and the worst possible outcome.

I have a name for AI-enabled democratised code:
Generative Inner Source.

It’s such an exciting time.

Let’s say that I asked you a question: “Who is the president?” — just those words with no other context. What would you say?

I reckon that the majority of you would say Donald Trump. But, the majority answer is directly tied to how generally well-known he is around the world.

If I was in France and asked you the same question, would you answer with Emmanuel Macron? Similarly, would the question being asked in Czechia be responded by Petr Pavel? Perhaps those answers are more likely, although the possibility of Donald Trump being an answer (given, again, his global exposure) is surely greater than zero each time.

Let’s get even more local. If we both worked in a large company which had a president, would you answer with the name of the company’s president, or would you think about it and perhaps answer with — again — a well-known political figure?

Tricky, isn't it?

It’s because in any AI data corpus, provenance and context are everything.

I’ll demonstrate with four image generators. Asking “an image of a president” produces the following responses.

Two of these are not surprising, but the other two perhaps are.

Google Gemini as an American AI service does a good job in creating a non-specific — though female — president. The iconography behind the president is undoubtedly based on the White House. Nothing really unexpected there.

Qwen is perhaps the biggest surprise. It’s a Chinese model and I was expecting Xi Jinping to be generated but it has instead, again, offered up a generic American president — though strongly reminiscent of Joe Biden.

Flux, which is a German AI provider, has offered the clearest real-life example: that’s very much Trump. It’s not Frank-Walter Steinmeier, the current German president, or indeed either the current Chancellor (Merz) or perhaps more famous Chancellors in recent history (Merkel).

Auraflow from San Francisco offers Donald Trump again, although it looks like a rather poor waxwork recreation.

Sometimes, giving simple but meaningful prompts like “an image of a president” offer up these interesting responses. They give an insight into both the provenance and context of the ingested data, and how the model was trained. Perhaps with the non-US models, they simply ingested more photos of Trump than any other president, because he has a greater global recognition and thus there are more images of him by volume. Or, the fine tuning offered up a context of “general use” and, again, an American president is — it’s assumed — going to be the most prompted version of one.

What this means when that if you are using AI in the enterprise, context in both data and usage is everything. If it’s really important that “president” in your enterprise means someone specific, then the data and training need to be geared to reflect that. Of course, in any language, words have multiple meanings, and thus the context of usage really needs to determine how you attune the models to suit. “Focus” might mean something very different if your model is used at Ford, or a camera manufacturer, or an opticians. There are, of course, many further words that go beyond such boundaries, where those words mean one thing in one language and/or culture, and something entirely different in another. “Chips” is an obvious example for those of us in the UK.

So, if you really want meaning, ensure that you have control over both the source (data) and the contexts by which that source is used. Otherwise, you’ll end up with your users in a bit of a mess — to coin a phrase, divided by a common language.

Who thought, decades ago, that Fred Astaire and Ginger Rogers would end up becoming AI strategists…?

So, let’s make environment management nice and simple, through building an app and making it available from Mac Apps in Finder and even the Dock if you wish.

First off… in Terminal, clone the wonderful Python Virtual Environment Manager.

git clone https://github.com/realcgslav/python-venv-manager-gui.git
cd python-venv-manager-gui

Then (yes, I know that this sounds ironic) create a virtual environment. NB you may need to replace python with python3 and pip with pip3, depending on what works on your machine.

python -m venv venv

Through Pip, install PyQt6, which is the Python-friendly version of the QT cross-platform GUI application framework.

pip install PyQt6

Run it.

python main.py

You’ll then see Python Virtual Environment Manager run, looking like this.

Nice, eh?

You can, of course, manage things from here but let’s make things easier and turn it into a Mac application.

Staying in the same directory:

pip3 install pyinstaller

PyInstaller turns Python scripts into executable Mac applications, invoking the Terminal (almost) invisibly to run them.

Again, same dir:

pyinstaller --onefile main.py

An executable Mac app will be created in the dist subdirectory. You can drag that to your Mac applications folder and rename if you wish.

And… you’re done. A quick and easy way to manage Python environments, from one place, without having to even touch the Terminal to do so. <thumbs up emoji copied and pasted several times over>

In amongst the expected hilarity, Sellers does an impression of Michael Caine, which you can see here at around 0:38.

It was slightly paraphrased into “Not a lot of people know that”. It became a famous saying by Caine, even though it was never said publicly by Caine himself.

30 years later, on the same show, Caine talks about the saying.

It has been used in countless impressions of Michael Caine, such as this one by Paul Whitehouse, in character as nosy neighbour Michael Paine.

Such is the ubiquity of Caine’s supposed saying that if you asked anyone — certainly, in the UK — to do an impersonation of the star, they would probably give you exactly those words.

What that person is saying is giving a set of words which one would associate with the originator, whether they are true or not. Voice + vocal tics + style of vocal delivery + sayings mimiced accent = an impersonation.

AI understands this too. The aforementioned saying is given by Anthropic’s Claude Sonnet 4 LLM, in the style of Michael Caine.

Sky News reporter Sam Coates experienced ChatGPT giving a similar type of response recently, when after prompting to give a transcript of Coates’ podcast, ChatGPT hallucinated and gave him a made-up transcript that, while looking like a de facto podcast transcript, had never actually been recorded or captured.

The responses to this video are interesting in themselves. Some express horror while others express surprise along the lines of “some people don’t know how to use AI properly”.

French philosopher Denis Diderot wrote in Paradoxe sur le comédien (1773) that actors must possess a judgment and penetration without what he called a “sensibility” — experiencing the emotions they are portraying while acting in-character and on-stage. Now, one might argue that the greatest actors must experience those emotions when performing in order to most effectively portray them, but in Diderot’s mind, the latent space between performer and audience is clear. While the audience may experience given emotions, the actor may not necessarily be “living” them.

A manifestation of this theory is impersonation, and this is where Diderot’s thinking is perfectly valid. The expression of another person doesn’t make it that person, and the audience knows that it’s not that person anyway, but a likeness. Il costume di Casa (1973) by Umberto Eco updates Diderot’s thinking to a consumerist setting — for example, the mass entertainment derived from going to a wax museum, and Jean Baudrillard in Simulacra and Simulation (1981) takes a more fundamental view, in that a photograph of (say) an apple is a known representative object without it being the object in the first place. The photo is not the apple.

The typical consumer AI interface challenges these theories, because they challenge our historically obvious understanding of what’s real and what’s not real. The hallucinatory output is mixed with factual/real content, and made to “look” real. However, because consumer interfaces both prevent us from tuning the hallucinatory factors — the “temperature” up or down - and they don’t highlight what is real and what isn’t in the output, we have to work it out for ourselves. And, that’s a challenge for a society whose media literacy has struggled to progress beyond “It’s on the Internet therefore it must be true”. We either have to start from a point of nothing being real/factual or everything being real/factual, because our brains can’t cope with the speed of addressing any sort of provenance while the AI interface is chugging out responses to our prompts.

There are no easy answers here, because it’s impossible to tell people to either believe or disbelieve everything. Madness lies either way. However, more informed user experiences can help us to understand this new, and fundamentally challenging environment in terms of how we prompt, process, and deal with real things that are impersonations, and impersonations that are real things.

In other words, we need a lot of people to know that.

Vibe coding is an interesting phenomenon. Devised by Slovak-Canadian AI expert Andrej Karpathy, the definition is to lose oneself in the production of a product, without necessarily worrying too much about the code — because AI code production services can take care of much of that for you. I have written code on and off throughout my career so as something of a dabbler, it was a concept that I found interesting.

I know that Electron is a framework that allows web-based content to be displayed within an executable application, and thus figured that using Electron was going to be the quickest way to do something.

So, I asked Watson Code Assistant.

I am making a website viewer using Electron. Generate the code that turns this into a MacOS app.

The response:

To turn your Electron-based website viewer into a MacOS app, you’ll need to package it using Electron Forge. Here’s a simplified guide:

Install Electron Forge: If you haven’t already, install Electron Forge globally using npm (Node Package Manager). Run the following command in your terminal:

npm install --global electron-forge

Initialize your project: Navigate to your project directory in the terminal and initialize your project with Electron Forge. Run the following command:

npx create-electron-app my-app

Replace “my-app” with the name of your project.

Configure your project: Electron Forge will guide you through configuring your project. You’ll be asked to choose a template (choose “None” if you’re starting from an existing project), and then you’ll be asked to configure your project’s metadata (like name, product name, etc.).

Build your app: Once your project is configured, you can build your app for MacOS. Run the following command:

npx electron-forge make

Find your app: After the build process is complete, you’ll find your app in the
out/my-app-darwin-x64 directory (replace “my-app” with your project’s name). This is your MacOS app.
Remember, this is a simplified guide. The actual process might involve more steps, depending on your project’s complexity.

Also, note that distributing apps outside of the Mac App Store requires a special entitlement, and Apple has strict guidelines for what can and can’t be distributed this way.

Lastly, always ensure you respect user privacy and comply with Apple’s App Store Review Guidelines.

I went through the instructions and, sure enough, made my first executable Mac app with some pre-installed dummy content:

It works! Cue minor sense of achievement etc etc.

So, if it works, what should I do with it?

I then had the idea of making a little player for IBM Community Radio. Because IBMCR uses a streaming format that is incompatible with Safari, you need to use Edge or Chrome to listen to it on a Mac. Could my Electron app act as as a radio player without the need to install a new browser?

Electron is simply a web browser under the bonnet of an executable file, and thus includes a web browser in itself (if you see what I mean). That browser is Chromium — the open-source variant of Google’s commercialised Chrome. So, by a stroke of luck, I then realised that my app could indeed play IBMCR, because it had Chromium inside it.

From that point on, it became rather easy to get the app to render IBMCR. Electron gives you all of the components necessary to make the web content inside the app. Simply replace the contents inside the src/index.html file to whatever you like. Here, I used the source of the IBMCR player, tweaked the HTML a little (eg absolute paths) for it to render correctly, and it worked.

The vanilla Electron app opens with a set of dev tools pre-enabled — basically the Inspect Element part of Chromium. That can be disabled, and it’s also possible to change the size of the app window. I worked out the player’s dimensions, and made all of those changes in src/index.js.

Building the app is simply a case of running the command in the requisite directory. It’s worth clarifying at this stage that I moved from electron-forge to electron-builder as I had a couple of system issues with electron-forge. Everything is fundamentally the same, however, in terms of building a little app.

There are online instructions to add an app logo of your choice — that’s an Apple standard.

The output is an executable Mac app, published in different ways. You get it as a .dmg (installable application), .app (run-anywhere application) and both as a .zip. You also get a blockmap file, which is a gzipped JSON file used for tracking updates, for incremental build purposes.

Here’s the app playing my show :-)

The app’s footprint is 252MB, which is of course really chunky as a simple app but it does have an entire web browser inside it, and it’s perhaps 10–20% the size of running a “normal” web browser anyway. If I was building a radio player from scratch then of course I wouldn’t do it this way, but as a demonstration of some AI-enabled vibe coding, it has worked out fine.

A huge advantage is that vibe coding writes a lot of the code for you, and a huge disadvantage is that vibe coding writes a lot of the code for you. As a “developer dabbler”, I know enough to produce the above and probably something more advanced. But, AI doesn’t necessarily know the full extent of your organisation’s code standards, security policies, packaging requirements, and so on. There’s a danger that it could do something at best bloated and suboptimal, and at worst, really dangerous and full of security holes. The trick here is to foster an organisational inner source culture which permits — ideally encourages — vibe coding within given guardrails and policies, but places very strong technical and practical limits on what is and what isn’t possible.

The brilliant Kaleidoscope television archive has been busy in the past few days, uploading reels of UK television advertising from the past 70 years. Of course, many of them are fascinating. Along with the videos (via YouTube), I have listed all of them below and commented on the company behind the ad and/or the ad itself where relevant.

I found them fascinating for all sorts of reasons (particularly the cars!)… although please bear in mind that many of them, in terms of their portrayal of particular people and social types, are very much “of their time” and caution is advised.

1950s/60s A

0:00 Gibbs SR
Unilever — the first ad that ever appeared on UK TV. It was broadcast on the first day of ITV, 22 September 1955, with the Associated-Rediffusion franchise broadcasting it at 8.12pm, just under 1 hour after broadcasting began

1:16 Lyril soap
Unilever — still available as Liril in India

2:14 British Insurance Association

2:46 Dubonnet
This became “Dubonnet Red” in the 1970s and remains an aperitif on sale in the UK

3:23 Rael Brook shirts

3:55 Schweppes drinks

4:25 Daz white knights
Procter & Gamble. This is quite the ad

5:32 Bank accounts

6:18 Esso
Warning: this ad features cartoon racial stereotyping

7:21 Eggs

7:53 Smiths Crisps

8:25 Mother’s Pride
Rank Hovis McDougall (since acquired by Premier Foods) — still available but now eclipsed by Kingsmill and supermarket bread. The Rank in RHM is Joseph Rank, who owned the Baltic flour mill in Gateshead (now the Baltic contemporary art gallery) and is a cousin of J. Arthur Rank

8:58 Shell
A classic ad here, with Bing Crosby

9:58 Sarongster girdles (!)

10:16 Eggs (Tony Hancock / Patricia Hayes)
See below

British Egg Marketing Board (1958–61)

Bernard Miles plays a Herefordshire farmer and the Lion Mark is represented by a cute little cartoon lion. This is a well-known campaign, featuring the famous “go to work on an egg” slogan, written by Fay Weldon. As per above, Bernard Miles was joined by Tony Hancock and Patricia Hayes for multiple campaigns between 1965 and 1971.

When UK egg production in the 1980s hit a crisis through mass Salmonella infection, comedian Jasper Carrott made the joke “go to work on an egg, come home in an ambulance”

1950s/60s B

0:00 Wool Marketing Board

0:49 Esso as above
See content warning above.
Exxon in the US. Esso was chosen as the UK brand as shorthand for “SO”, the initials of Exxon’s predecessor Standard Oil

1:59 British Insurance Association as above

2:20 Rael Brook shirts as above

2:53 Bank accounts as above

3:39 Bird’s Eye
Unilever; frozen food business later sold to Nomad Foods

4:42 Bio-Tex (Thora Hird)
Unilever; product is still available in Scandinavia. This campaign was used as a question in Shooting Stars. The reason why Thora stays still at the end is to add on some run-out time

5:15 Hoover appliances

5:45 GPO (Irene Handl)
From the days when, as with most other European countries, post, telegraph, and telephones were bundled into one state-owned organisation (“PTT”)

6:17 Meccano
Very much the traditional ad here, about a boy becoming a man through learning engineering

7:03 Parker pens

7:36 Watney Pale Ale (Scaffold)
Changing the lyrics to “Lily The Pink”. The business was bought by Grand Metropolitan (now Diageo) and was closed; its London brewery in Mortlake was taken over by Anheuser-Busch to brew Budweiser and closed in 2015

8:06 Eggs as above

8:38 Gibbs SR toothpaste as above

9:41, 10:42, 11:44 Eggs (Hancock/Hayes as above)

12:46 Lux soap (Diana Rigg)
Unilever

13:30 Frigidaire appliances
In the race to build diversified conglomerates over the 20th century, Frigidiaire — rather bizarrely — became part of General Motors (this is not an isolated relationship; British Leyland also owned a refrigeration business). The brand is now owned by Electrolux

14:17 Shippam’s Paste sandwich spread

15:20 Kleenex tissues
Bowater, a packaging conglomerate and the UK licencee for Kleenex, and since acquired by Kimberley-Clark. This is a really interesting ad stylistically as it is live-action, but bookended by animation

16:19 Hotpoint refrigerators
Associated Electrical Industries, which eventually became part of huge industrial conglomerate GEC (which itself spectactularly collapsed in the 1990s); Hotpoint is now owned by Whirlpool

17:20 Capstan cigarettes

17:50 Player’s cigarettes

18:54 Cadbury’s Flake
From the dominant Cadbury’s business in the UK, which became part of the food, drinks, and homecare conglomerate Cadbury-Schweppes; later demerged and acquired by what is now Mondelez

19:22 Oxydol washing powder
Procter & Gamble, still available in the US

20:25 Flash floor cleaner (Mrs. Flash)
Procter & Gamble. Again, another rather splendid ad in the style of Daz above in, what I think is, Newark-upon-Trent

21:26 Alberto VO5 hair spray
Alberto-Culver, now part of Elida

21:58 Fairy Snow washing powder
Procter & Gamble

1970s

0:00 Hotpoint

1:24 Campbell’s Soup

2:26 Croft Original wine
An early airing for the brilliant slogan “One instinctively knows when something is right”

3:29 Ravenhead glassware

1972

0:00 MacFisheries Dutch campaign
Unilever, when they used to own retail businesses including this fish chain (which became the Mac Market supermarket, as well as Nordsee in central Europe). You have to watch this, just for the sweet young lady desperately trying to stay still at the end to give some run-out time

0:53 Mace supermarkets

1:24 Pricerite supermarkets

1:54 Tesco supermarkets

2:03 Woolworth — The Andy Williams Album

2:35 Rael Brook shirts

1973 — Car ads

Many of these ads have the same telephone response line: 01 388 1122, set up in Thames Television’s Euston offices. Here’s the background.

0:00 Mercedes / Denham Service

0:48 Toyota

1:32 Toyota

2:33 Bristol Street Motors
Portable TV offer “Gets all 3 channels”

3:27 Opel Rekord
Opel was withdrawn by General Motors in the UK in the 80s, in order to focus on its Vauxhall brand

4:52 Ford / Strakers

5:52 DAF
No longer a car manufacturer; car business bought by Volvo

6:59 BMW

7:43 Renault 6

8:29 Ford Capri

9:51 Ford Capri
A young Geoffrey Palmer in what I think is Marylebone railway station

11:26 VW 1303
This has a brilliant portrayal of London kids

12:13 Chrysler / Hillman Avenger
Hillman’s owner Rootes was bought by US manufacturer Chrysler, and rebranded accordingly (hence the emerging references to the parent company). In 1978, Chrysler bailed out and sold the entire European operation to PSA (owners of Peugeot and Citroen). Eventually, Chrysler, Peugeot, Citroen, the aforementioned Opel and the later-mentioned Fiat and Vauxhall all became part of one company, Stellantis

12:44 Chrysler 2L

13:15 Chrysler / Hillman Hunter

13:45 Chrysler / Hillman Avenger

14:17 Chrysler range

14:47, 15:18 DAF

15:26 Toyota

16:32 VW

17:17 Toyota Corona
Starring a young Colin Baker

18:04 Austin Allegro
Starring a young Bill Treacher, aka Arthur Fowler

19:04 Vauxhall
This is a well-known ad, simply because it’s so weird

20:08, 21:30, 22:52 Audi 80
Very odd, vague slogan here — a car that is “3 to 4 years ahead of its time”. This 80 is only the second Audi to be developed under VW ownership after its purchase from Daimler-Benz

21:11, 22:33, 23:55 Audi 80 / Smith and Gardner Liverpool

24:32, 25:56 Datsun
Now, of course, Nissan; Datsun was relaunched a few years ago as Nissan’s budget brand for emerging markets, but it wasn’t successful and was later dropped

27:19, 28:14, 29:02, 29:54, 30:44 Fiat
This is fascinating. Fiat had a major issue with body rust, and this tackles the issue head-on but also invites the question “what about the cars that didn’t have the new body treatment”

1981

0:00 Haywards pickles

2:14 Crosse and Blackwell cooking ingredients
Owned by Nestle for a long time, since divested to different businesses in each global territory

3:36 Black Magic chocolates
Rowntree Mackintosh, now part of Nestle

4:45 Warwick Records — Adam Faith

6:10 Warwick Records — Brotherhood of Man

7:31 Harrods

I spent the summer of 1995 in California. Between the second and third years of University, I got up and went on my own. The experience was unquestionably life-changing; meeting new people, engaging in new experiences, and even helping to move someone from San Diego to San Francisco (and Sarah, I’m still sorry that I accidentally packed your mom’s broom).

I didn’t understand at the time that SF was going to be the epicentre of consumer technology, as we now understand it. I do remember meeting a bunch of developers, who let me stay at theirs in the Mission, and had broadband at home — a “fixed line” at home that I found extraordinary. But, even as this little world became part of everyone’s universe, “tech” wasn’t new for San Francisco and the valley. It had already borne global companies such as Hewlett-Packard and Apple but of course the situation these days is much bigger — in every way.

The US builds brands like no other country on earth. You don’t eat a burger and fries, you have a McDonald’s. You don’t watch streaming TV programmes, you Netflix and chill. You don’t drink coffee, you have a Starbucks. The ability of US companies to make their brands stick in such multicontextual ways is really unlike anything else, anywhere else.

But, what is often missed is not just the ability of the US to build businesses and brands, but the ability to vertically-integrate them. Look at “tech” this way. You’re using this website or app (Medium, from the US) on an operating system (probably from the US) on some hardware that is most likely to come from the US. And, when you’re reading about “tech”, you’re doing it through browsing US publications which largely cover US tech companies. Facebook / Meta, Apple, Uber, and Google all grew massively in the 2010s because we were all reading TechCrunch. It’s essentially a media-technology complex, that builds a virtuous circle for US tech hosting US media to talk about US tech.

And, it‘s this media-technology complex that is becoming increasingly globalised. National broadcasters across Europe, all of whom have been far too slow to get this stuff, are being totally outclassed by the big-budget US streamers (Netflix, Prime) and the small-budget homegrown shows on YouTube. It globalises US culture in a way that, again, doesn‘t happen elsewhere. I am eternally grateful for it myself, but it seems rather pathetic that with all of its many cultures, Europe cannot catch the US in terms of exposing and commercialising what it has.

To be clear, I’m not criticising the US at all in this article. The ability for its companies to execute, as per the examples above, is astonishing. It’s a beautiful, diverse, unique country, I love every minute that I can spend there, and I proudly work for an American tech company.

But, because it’s so good at all of these things, it makes the socio-economic and technological challenge for Europe, that much harder. China vertically integrates (economy / business / technology / media) politically, while the US vertically integrates economically. Both vertically integrate socially. Europe doesn‘t do either at all well.

The fundamental problem that Europe has, is: what is it? Starting a tech business in the US usually requires the adoption of just one language. Once the business is successful, it can then branch out to other languages as per geographic adoption. European businesses in Italy would start with Italian, those in Sweden with Swedish, and so on. The EU has 24 official languages, so building an European audience is nowhere near as easy as in the US. That can’t be “fixed”; it’s just a fact. Starting a business in Europe requires careful consideration of the user base across multiple countries and jurisdictions, and investing in legal and translation support in a way that an English-first big single country as the US simply doesn’t need to do. That becomes exponentially harder when conversational AI is part of your offering.

Let’s stick with AI for a moment.

If you were starting an AI business, where would you do that in Europe? There are clearly some obvious choices, but don’t you also want to keep your business in a smaller country if you’re from there? Again, the single big country of the US simply doesn’t have this problem. To up sticks from (say) Bucharest and move to Berlin is not easy. It’s not impossible either, but it makes the unicorn status of Revolut and Wise all the more heroic, given that their founders moved to from Ukraine and Estonia to London to grow their businesses.

However, even the nascent sector of consumer AI looks like it’s dying in Europe. Mistral is the only big European player, but it already appears to be struggling. This is a massive reputational loss to a continent that is already seen as being a frustrating and frankly poor place to do business. Of course, there are some successes – Revolut, Wise, Klarna, Vinted – but there simply isn‘t enough of them to say that Europe as a whole has a successful tech sector when judged in terms of economic success.

Europe is basically a big bag of complexities — some are easier to overcome than others. Some national and transnational governments like the UK and EU are starting to finally get it, but I get the feeling that — particularly with the EU’s actions — that it’s going to boil down to “soft interventions” rather than the harder, faster policies required to regain the technological competitiveness so urgently needed.

Here are some thoughts as to how that could, at least partially, be achieved.

Pan-European location strategy

Cluster businesses in particular cities and regions, and support people and businesses moving there to grow, with support available to overcome cultural and linguistic barriers, and to internationalise products.

The fintech scene in London has been critical for Revolut and Wise. Berlin has a great ethical tech scene. What about graphics, or connectivity, or medical, or AI / AGI, or futures research? Unfortunately clustering such areas will piss off some other European countries, but it’s for the greater good, given that a company with (say) a Romanian CEO locating his fintech business in London will most likely grow sufficiently to open a future office back in Bucharest.

Single European company registration

The SE model has been great for large companies, particularly German ones, to grow across Europe from one corporate entity. Now do this for SMEs. Make it low/zero tax, zero paperwork, and allow such an entity to operate in any European country. Basically take what Estonia has done to create a pseudo-corporate entity for foreigners, and make it fully legally recognised. This can be managed by the EU but the UK and EFTA countries should be part of it (essentially facilitating the single market through digital means). Thankfully, it‘s an idea that is already likely to be adopted.

Open source

Encourage European businesses to foster open source into their working practices. Open is good in every sense; it delivers innovation, builds community, and creates social value.

One of the biggest open source success stories ever comes from Europe: Linux. Open source should be celebrated, reinforced, and promoted as a way to do business.

Social impact

… and on the point of social value, build companies that have a lasting social value. This is where Europe can really differentiate. Software may be eating the world, but it should also give back. How can European business think and act in a way that is, of course, about building wealth, but also about respect, equality, and environmental stewardship? There must be structures in place to allow for that, in a way that is totally non-disruptive to business growth.

Keep the backbone

Regulate “big tech”, sure, but don’t let such regulation get in the way of European business growth. As US consumer tech companies become increasingly politicised, the EU needs to more convincingly argue a case for such regulation and why it is necessary (given that many consumers need educating on their rights in the first place, even before they understand the reasons for regulation).

Don’t stop doing what you’re good at

If we move beyond “tech” as being short for contemporary communications technology, then Europe actually has some great technology companies: Siemens, SAP, Dassault, Amadeus, Ericsson, Deutsche Telekom, and Schneider.

Similarly, we are great at retail (Inditex, Carrefour, H&M); pharma and bioscience (GSK, Roche, Bayer, AstraZeneca); and commodities (Glencore, Vitol). However, many of these companies are risk-averse and yet yearn to play a part in future technologies. They shouldn’t stop doing what they do, but they should definitely use more of their financial power to invest in startups and support ecosystems.

Create the media-technology complex

As per my earlier point about the US, Europe needs to be much more vertically integrated. This is a golden time to build media businesses which both talk about and support the businesses that need both growth and coverage. European media is so far behind and it hasn‘t understood the part that it could, and should, play in reinforcing its own culture. Startups need exposure and there is no obvious way of doing that in Europe right now. Culture and media is so important to this, and it seems to be tragically forgotten.

I cannot emphasise this point enough. Create spaces for startups, technology companies, and tech in general to be talked about. Europe is fucking pathetic at it.

Build funding networks

Companies need funding, from traditional and non-traditional sources. Build those connections. Be risk-aware, not just risk-averse by default. This is the most challenging point, because while Europe is actually quite good at Private Equity (for example), the money tends to flow into other sectors which are seen as less risky. Encourage founders to connect in order to jointly build new PE companies off the back of successful maturing businesses.

More generally, funders need to be quicker, and better. ElevenLabs is a Polish AI voice startup, already valued at $3.3bn, with funders including A16z, Salesforce, and Sequoia. Why aren’t European funders more involved in deals such as this?

Learn

Quickly learn from failure. Why are European tech companies going bust, or selling out? What are the regulatory, economic and cultural hurdles to prevent that from happening again? If any of the aforementioned suggestions don’t work quickly, then pivot the fuck out of them. Don’t do yet another “3-year programme”-type intervention which is lovely in theory but the world has moved on by the end of it.

A summary of the approach is: be fast and pivot quickly, within the context of both greater risk on the part of funders, and less risk on the part of society. Build your own identity, not trying to be the fucking “Silicon Valley of Europe” every time.

The old slogan for MG cars describes this approach in just two words:
Safety fast.

For more on this topic and to get involved in building Europe’s digital ecosystem, visit the European Accelerationism campaign site.

Crack German: 5/10
It has been 7 years since I publicly started and wanted to be at B1 level by the end of the year. I’m not there yet.

Consolidate career: 6/10
A little better in 2024 but still work to do, but as I’m taking on a new role at IBM in 2025, I shall give that all of the energy and enthusiasm that I have. I’m genuinely really looking forward to it; more info to follow.

No nail biting or picking: 10/10
Conquered.
Gone from 1/10 last year to 10/10 this year. I suffer from flu quite a lot and thought through how to avoid bacteria. From that point onward, I stopped biting my nails. This is a major personal success.

No oversleeping: 3/10
Fitness: 8/10
Same as in 2023 and the two are inextricably linked.

No inferiority complex: 6/10
Not bad, but needs more work (not sure that I can ever get to 10/10).

Coaching: 4/10
Thanks to everyone that has been a coachee this year; I aim to do much more coaching in the new year and finally get the accreditation that seems so elusive.

Help Ukraine: 1/10
I didn’t, and Ukraine needs more help than ever. A priority for 2025 as the war approaches an inflection point.

Art and culture: 1/10
More next year.

Something fun/productive: 2/10
Nothing this year although I have an idea that I really want to start working on in 2025.

Get out more: 7/10
Experience more is what I put last year and I did. There’s more to do — literally — and I hope to spend more time with others in 2025. For the new year, I would like to wrap up my intentions in two words: embrace euphoria. Enjoy life even as the world is getting crazier.

I’ll add a new one for 2025: read more.

I’ll finish with my favourite album and song of 2024. The song is Stačí by Supa, Vec, and Kriss Krimm. It’s a truly extraordinary slice of rap, in Slovak.

The album is Short and sweet by Sabrina Carpenter. I listen to a lot of music each year — and mix much of it, too — and although I take in a lot of complex and often quite brutal music, this album is a cracker. It’s superbly produced and the lyrics are <chef’s kiss>.

Thanks for reading any of my posts this year.
May the 2025 of you and yours be one of love, peace, the best of health, and the product of everything you wish for.

Over the past year, I have been sharing what I’d like my next “fun little thing” to be. I’m pleased to tell you what that is: a publishing venture covering less-well-known technology stories from across Europe, launching in 2025.

Books on, from, and about Silicon Valley are plentiful. It’s easy to find a book covering business models, leaders, or company stories from that area. Silicon Valley’s place in the history of communications technology is assured. But, there are so many additional stories from Europe that are less well-known globally.

The concept of Mikroverlag is to cover specific histories related to European technology: its innovations, its people, and also, its failures. Central and Eastern Europe is particularly ripe for untapping as there are many stories of government-led and people-led innovations that are in danger of being lost.

That said, the point of Mikroverlag is to cover technology stories from the whole of Europe. We might cover, for example, the development of Minitel in France, or the story of the Sinclair ZX Spectrum clones in Communist-era Czechoslovakia. The first digital computer, the German Z3, might be a future topic, or perhaps a biography of Benoit Mandelbrot.

I’m just making these up. This is where you come in.

If there’s a topic that you would like to see us publish, or you have an idea for a book that you would like to write, then please get in touch: mikroverlag@perininetworks.com.

I have absolutely no idea how to get this off the ground. We’ll all learn as we go. It’s really a bit of fun, but if we’re successful, we’ll unlock some of these important but untold stories.

Paul