of all web traffic came from machines, not people.1
of newly published online articles are primarily AI-generated.2
of people click a link when Google shows an AI summary. Without one, 15 %. On the sources the summary cites: 1 %.3
When machines do the talking, the human becomes precious.
What is happening
Ice was a luxury until freezers arrived. Then cold became ordinary and the rare thing was a glacier.
Content has become infinite. What is running short is trust, authenticity and lived experience. Anything a machine can produce in seconds stops being an argument.
A peacock’s tail is a burden. That is exactly why it is honest: only a healthy bird can afford to drag it around.
Biologists call this the handicap principle. A message that costs nothing says nothing. Showing up in person, filming on location, signing your name — all of it costs. That is what makes it credible.
A print of a painting hangs in a million living rooms. People still queue to stand in front of the original.
Walter Benjamin called it the aura: the here and now of a thing, its presence in a particular place at a particular time. Copies multiply. The original does not.
AI in the engine room. People at the front desk.
„The medium is the message.“
When content becomes infinite, scarcity shifts to trust, cost and originality.
Service 01
Everything your business writes is now read by machines first — by search engines, answer systems, procurement platforms and increasingly by your own people’s AI assistants. We make sure what you meant is what arrives. Outward, so you appear in answers. Inward, so your own knowledge stays findable.
We sit in on the actual work for a day. Not a questionnaire: the real process, with the shortcuts and the annoyances nobody writes down.
One narrow case, built small, measured honestly. If it does not hold up, we say so and stop. That is cheaper for everyone than a year of enthusiasm.
Into your accounts, on your access, with your data. Documented in a way your people can read.
Systems drift. Models change, interfaces move, processes grow. Somebody has to keep watching — either us or you, but somebody.
Since 2 August 2026, EU rules require AI-generated content to be recognisable as such. We help you work out what that means for your own content and how to label it without turning every page into a legal notice. Binding legal questions go to lawyers, not to us.
Most AI projects fail on the process, not the technology, so we start with the process.
Service 02
Models now learn from their own output. What disappears first is the rare, the local, the specific — exactly what sets your business apart from every other one. We film, photograph and write on location: with the people who actually do the work, in the place where it is done. This is not a counter-proposal to AI. It is the material without which AI cannot say anything true about you.
None of this proves what happened in front of the lens. It proves where a file came from and who agreed to what. That is all the technology does, and all we claim. On top of that, many platforms strip this data on upload, so the chain can break outside our hands. We document anyway — a visibly broken chain is worth more than no chain at all.
What is rare, local and specific cannot be invented by a model. It has to be recorded.
Service 03
Two ways in. Pick the one that fits how you want to work afterwards — that decision matters more than the choice of tool.
We build the process and operate it. You get the result, not the maintenance.
We build the first one with your people watching, then hand over everything — the prompts, the tools, the reasoning.
Automate the cost, not the relationship.
Bill Gates put it roughly like this: automation applied to an efficient operation magnifies the efficiency; applied to an inefficient operation, it magnifies the inefficiency. Which is why we start with the process, not the tool.
„The hottest new programming language is English.“
Automating a broken process makes it break faster, so the process comes first.
Service 04
Most businesses hold more data than they think and less usable data than they hope. The same company spelled three ways, fields someone meant differently in 2019, numbers without a unit. We clean that up, reconcile it and analyse afterwards — in that order, because analysis on messy data is not half right, it is wrong to several decimal places.
Turn records that simply grew into one you can cite.
Two systems that ought to know the same thing rarely do — holding them against each other is work of its own, not a by-product of the clean-up.
Answer the questions nobody could ask before.
Analysis on messy data is not imprecise but wrong — and more dangerous than none, because it looks convincing.
Tools
Not a logo wall — the toolbox that is actually open on a Monday. Which of these gets used is decided by the process, not the other way round.
Claude · Claude Code · Cursor · Gemini · Model Context Protocol (MCP)
TypeScript · React · Next.js · Node.js · Python · Git · GitHub Actions · Docker · REST APIs and webhooks · Vercel · edge functions
n8n · Make · webhooks and APIs
dataCycle · Supabase · structured data to schema.org · JSON-LD · XML sitemap and robots.txt · Open Graph
ClickUp · Google Workspace · Gmail · Calendar · Drive and Docs · Sheets · Dropbox
WordPress · statically generated pages · Canva
Everything runs in your accounts, on your access. What we set up belongs to you — including after we are gone.
For AI teams
Licensed video and audio produced by a film crew, with documented provenance, signed consent covering AI training, and cleared rights. Commissioned to your spec, or from unreleased archive.
Shumailov et al. showed in Nature that generative models trained recursively on generated data lose the tails of the original distribution — the rare cases disappear first, and the output flattens toward the average. Real human material is not a nice-to-have. It is the correction.4
The archive of everything already written and filmed is being bought up. What nobody has finished scanning is the present: new places, new faces, work being done with hands, languages still being spoken.
Michael Polanyi called it tacit knowledge: we can know more than we can tell. A cook adjusts by feel. A technician hears the fault before the meter shows it. That knowledge exists only where it is performed — which is to say, on camera or nowhere.
Reddit licenses its conversations to Google and OpenAI. Bloomberg reported in January 2025 that AI companies were paying one to four dollars per minute for unreleased video, with premiums for 4K, drone footage and 3D animation.6
New material, shot to your specification by a film crew.
Unreleased material from the network and its partners, prepared and cleared.
Where we are careful: C2PA proves provenance, not truth. It tells you where a file came from and what was done to it — it cannot tell you whether what was filmed was staged. And many platforms still strip the metadata on upload, so the chain can break outside our hands. We document anyway, because a broken chain you can see beats no chain at all.
„We can know more than we can tell.“
AI needs material it cannot generate itself, with provenance and consent that hold up.
Proof of concept
Behind data.alive.family is a network of creative people built up over years around Lake Constance and beyond: filmmakers, camera operators, drone pilots, sound engineers, event producers, photographers and writers. Every project gets the team that fits it — not the team that happens to be free.
We did not start filming because AI needed data. We were already there.
Everything here was built for a business inside the network or for a partner, and used daily before it was offered to anyone else. Names appear only where we are allowed to use them.
Everything was tested in our own businesses first.
The network has been producing human content and running its own software for years.
How we work
data.alive.family is the AI, data and automation arm of the TeamAlive network: creatives, technicians, filmmakers, drone pilots, event people and business owners across the border region. Each project gets the people it actually needs.
The people here run restaurants, event technology, software and film production. The tools we recommend are the tools we use on our own Mondays.
GDPR-compliant by default, hosting in the EU, and a clear answer to where every piece of data sits. If a project needs something outside the EU, you hear it before it happens, not after.
Based in the Lake Constance region, working across Austria, Germany, Switzerland and beyond. Being from somewhere is an advantage when the job is to film it.
A network of practitioners who use the tools they recommend, with data kept in the EU.
Questions
AI-generated means the machine produced the content. AI-assisted means a person made it and the machine helped — correcting, shortening, suggesting. Under Article 50 of the EU AI Act, the first case must be marked. The second is exempt as long as the assistance does not substantially change the content or its meaning.
If you publish content that was generated or substantially altered by AI, yes, since 2 August 2026. If AI only helped you edit something you wrote, no. The grey zone in between is where most businesses actually sit, and it is worth going through your own cases once rather than guessing.
We do not publish prices, because the honest answer depends entirely on the process. What we can say: we start with one narrow case you can measure, not a platform. If that case does not pay for itself, there is no reason to continue, and we will say so.
That is one of the two ways we work, and the one we recommend when you have people who enjoy this. You get the prompts, the configurations and the reasoning behind them. It runs in your accounts. We stay reachable without being necessary.
Because models trained on model output degrade. A Nature study by Shumailov and colleagues showed that recursive training on generated data makes the rare cases disappear first and the output drift toward the average. New human material is the correction — and unlike the archive of the past, it is not finite.
Three layers. Content Credentials (C2PA) signed at the camera where the equipment supports it. A documented record of place, date, crew and every processing step. Signed releases from everyone appearing, with an explicit AI-training clause. None of it proves what happened in front of the lens — it proves where the file came from and who agreed to what.
No. We use AI for processes — the work behind the scenes. We do not use it to replace the people in front of the camera. That is the whole position, and it is also why Service 02 exists.
In the EU, by default, and we tell you which provider. If a specific project needs a service outside the EU, you hear about it while the decision is still open.
Check it yourself
Enter your address. We fetch the page once — without JavaScript, the way a crawler does — and show you what actually arrives. Takes a few seconds, costs nothing, and needs no email address.
Get in touch
No qualification form, no funnel. Describe the thing that keeps costing you a morning every week, and we will tell you honestly whether we can help.