The Next Data Layer Will Not Be Text. It Will Be Reality.
- Sven Franken
- May 7
- 5 min read
The Next Data Layer Will Not Be Text. It Will Be Reality.
For the last decade, artificial intelligence has been trained on the internet.
Text, images, videos, comments, reviews, clicks, profiles, code repositories and documents became the raw material of the AI race. The web became the mine. The model became the refinery. The frontier labs became the industrial powers of the new intelligence economy.
But that mine is reaching its limit.
The internet has largely been eaten. More content is still being produced every second, but for frontier AI development, “more” no longer means “better”. More posts. More synthetic noise. More SEO sludge. More generated summaries of generated summaries.
The next leap in intelligence will not come from scraping another billion web pages.
It will come from understanding the real world.
And that is where Augmented Reality becomes strategically urgent.
From web data to world data
The first wave of AI was built on language.
The second wave became multimodal: text, image, audio and video.
The third wave will be interactive.
Not just models that read the world, but models that understand what happens in the world. Models that learn from movement, attention, spatial relationships, physical context, object interaction, environmental signals and real-world feedback loops.
A web page tells you what someone wrote.
A search query tells you what someone wanted.
A profile tells you who someone might be.
But real-world spatial interaction tells you what someone actually does, where they do it, how they respond, what they ignore, what they approach, what they revisit and where the physical environment creates friction or opportunity.
That is not content data.
That is context data.
And context data is where the next AI advantage will live.
Brands are about to make the same mistake again
This is where the warning becomes serious.
Brands cannot afford to sleepwalk into the next version of the social media advertising trap.
The first trap was simple. Brands built audiences on platforms they did not own. They invested millions in content, communities and engagement, only to discover that access to their own audience could be throttled, priced, auctioned and resold back to them.
They became tenants on someone else’s infrastructure.
They paid to build attention.
Then they paid again to reach it.
Now many are preparing to repeat the same mistake with AI and spatial computing.
If brands wait until frontier labs, hyperscalers and platform owners dominate real-world interaction data, the result will be predictable. The data will be captured elsewhere. The intelligence layer will be owned elsewhere. The customer interface will be mediated elsewhere.
And brands will once again be forced to buy access to relationships they should have built themselves.
That is the next trap.
Not social media ads.
Spatial intelligence ads.
Context auctions.
Real-world behavioural targeting controlled by someone else.
If brands do not start building their own spatial data layer now, they will enter the next decade as data-poor customers of data-rich platforms.
Frontier labs need better feedback loops
Frontier labs such as DeepMind, OpenAI, Anthropic, Meta and Google are no longer only competing on compute. Chips, power, memory and datacenter capacity still matter enormously. But another bottleneck is emerging: data quality.
The next frontier is not simply more data. It is better feedback loops.
That means proprietary data, multimodal data, reinforcement learning, human feedback, tool-use traces, enterprise workflows, scientific simulations and real-world task completion.
This is especially relevant for DeepMind. Its historical strength has never been only text. It has been reinforcement learning, simulation, planning, games, scientific discovery and systems that learn through interaction.
The world is moving from models trained on what humans published to models trained on how reality behaves.
That shift is massive.
And most companies are still looking in the wrong direction.
Web profiling is yesterday’s game
There is still an entire industry obsessed with web-based profiling.
Cookies. Clickstreams. Behavioural targeting. Lookalike audiences. Retargeting funnels. Social engagement. Keyword intent. Page views. Heatmaps. Conversion pixels.
Useful? Yes.
Strategically decisive for the next AI era? No.
Because these systems mostly describe behaviour inside a digital surface.
They know what someone clicked, but not what someone physically experienced.
They know what someone searched, but not what someone saw in context.
They know the profile, but not the place.
And the future will belong to systems that understand place.
The next valuable data layer will come from the interaction between people, devices, objects, locations, buildings, cities, venues, products, exhibitions, events and physical infrastructure.
That is the layer where Augmented Reality lives.
AR is infrastructure
For too long, Augmented Reality has been misunderstood as a visual effect.
A floating 3D object. A filter. A game mechanic. A marketing stunt. A nice-to-have engagement layer.
That view is dangerously shallow.
AR is not only about placing digital content into physical space. Properly designed, AR turns physical space into an addressable, measurable and interactive digital environment.
A stadium becomes a data environment.
A museum becomes a data environment.
A city centre becomes a data environment.
A retail location becomes a data environment.
A festival becomes a data environment.
The value is not only in what the user sees.
The value is in what the system learns.
Where did the user stand? What did they look at? Which object triggered interaction? Which route did they take? Which physical context changed behaviour? Which environment created attention, conversion, learning, safety, emotion or memory?
That is the beginning of real-world intelligence.
TROPOS AR sits exactly at this shift
This is why TROPOS AR should not be understood merely as an AR SDK.
It is an infrastructure layer for the data of tomorrow.
The TROPOS AR SDK enables developers, brands, institutions, venues and cities to place digital experiences into the real world. But the deeper strategic value is that it creates the operational layer through which real-world context can become structured, addressable and usable.
Not just as entertainment.
Not just as marketing.
Not just as engagement.
But as a future data layer.
A layer where physical space becomes machine-readable.
A layer where human interaction in the real world becomes understandable.
A layer where spatial behaviour becomes part of the intelligence stack.
TROPOS AR is not chasing the old internet.
It is helping build the bridge to the post-web data layer.
Own the context, or rent it forever
The web created the attention economy.
Spatial computing will create the context economy.
In the attention economy, brands paid to capture the user’s gaze on a screen.
In the context economy, value will come from understanding the relationship between user, place, object, task and moment.
That is much more powerful.
It is also much more dangerous if brands enter it passively.
The companies that build the spatial layer will own the learning loop. The companies that merely advertise inside it will rent access. Again. At increasing cost. Under rules they do not control.
This is the moment for brands, venues, cities, sports clubs, museums, festivals and enterprises to act.
Not in five years.
Not when Apple, Google, Meta, OpenAI or another platform has already defined the rules.
Now.
Because once the spatial data layer is captured, the market will not politely reset itself.
For brands, the choice is brutally simple.
Build your own spatial data position now.
Or become a tenant in someone else’s reality layer later.
TROPOS AR is the key to that future.
Because the next frontier is not more text.
The next frontier is reality.
Comments