dm has always had one of the more interesting brand promises in European retail:
“Here I am human. Here I shop.”
That sentence is stronger than most corporate AI strategies. It is also more dangerous, because it gives us a real test.
Not: does the company use AI?
Not: does the AI make assortment, prices, logistics or availability sharper?
But: does the system still make people more capable?
That is the point where a New Work perspective becomes useful. Frithjof Bergmann's original New Work was never a wellness slogan for office culture. It was a hard question about the conditions under which people can do work that does not reduce them to replaceable functions.
AI brings that question back, but with better dashboards.
The generous reading
There is a credible, generous reading of AI in retail.
If AI helps a company reduce repetitive bureaucracy, improve availability, avoid waste, support employees with better information and free people for more meaningful customer interaction, then AI can strengthen a human-centered institution.
In that version, AI is not the replacement of the craft. It is the removal of some of the sludge around the craft.
Nobody becomes more human by manually reconciling badly structured spreadsheets at 17:43 because three systems did not agree with each other before lunch.
There is a good case for AI as relief.
But relief is not the same as dignity.
The harder reading
The CLAW-Fabric run we made on this question did not produce a clean victory for the optimistic thesis.
It passed the trust gate, worked through 996 source items, and still landed in a cautious plausibility band: 36-54%, center 45%.
In plain language: the claim is discussable, but not proven.
The strongest warning was not that dm, or any similar institution, secretly wants to replace people. That would be too simple.
The stronger warning is structural:
AI can compress work while preserving the language of human-centeredness.
It can redefine value as “effort saved”.
It can move judgment from the person in the situation to the system that scores, predicts, routes, recommends and optimizes.
And then everyone still says “the human remains central”, while the human becomes the final interface of a decision chain they no longer shape.
That is the polite version of alienation.
The employee is still there. The smile is still there. The brand sentence is still there.
But the room for judgment has quietly moved somewhere else.
The New Work test
From a New Work perspective, the question is not whether AI reduces costs.
Of course it will be used for that.
The question is whether the gains are reinvested into human capability or extracted as pure compression.
A human-centered AI strategy in retail would have to show at least five things:
1. More autonomy
Employees gain room to decide, not just faster instructions.
2. More competence
AI helps people understand products, customers, supply chains and situations better. It does not turn them into operators of invisible rules.
3. More relation
The customer encounter becomes more attentive, not more scripted.
4. Contestability
Workers can question, override and improve system decisions without being treated as friction.
5. Visible redistribution of efficiency gains
If AI saves time, where does that time go? Into staffing resilience, learning, better service, better schedules, less pressure? Or only into margin?
Without those conditions, “AI serves the human” remains a mood.
And moods do not govern systems.
The Slogan Is Not the Architecture
This is why the dm example is interesting beyond dm.
Every serious organization will soon claim that its AI strategy is human-centered. Hospitals will say it. Banks will say it. Retailers will say it. Public administrations will say it. A few will even put it into a tasteful PDF with a diagram that looks like accountability took a yoga class.
The real question is architecture.
Who defines the metric? Who sees the data? Who can object? Who benefits from the efficiency? Who carries the pressure when the optimization becomes normal?
If the answer is unclear, then the human is not at the center.
The human is in the marketing layer.
A Better Standard
The strongest version of a dm-like AI strategy would not be anti-technology.
It would say:
We use AI where it strengthens availability, orientation, learning and service.
We do not use AI where it reduces employees to compliance endpoints of a hidden optimization system.
We measure not only efficiency, but autonomy, competence, trust, quality of interaction and the ability to object.
We publish the rules-in-use.
We treat human judgment as infrastructure, not decoration.
That would be a serious standard.
It would also be good for business, but not because it sounds nice. Because retail is not only logistics. It is contact. Memory. Local trust. The small human moments that make a store more than a warehouse with lighting.
AI can support that.
It can also flatten it.
The difference will not be decided by the slogan.
It will be decided by whether people inside the organization become more capable, or merely more optimized.
Further reading
Flowbook KI tracks this broader line under trust architecture, freedom substrates and New Work in the age of AI:
https://flowbook.newwork-newculture.dev/
CLAW-Fabric note:
`DMMENSCH01-20260708T150311Z-ba725c`
Plausibility band: 36-54%, center 45%
Trust gate: passed
Main warning: human-centered rhetoric can coexist with structural compression unless autonomy, contestability and redistribution are explicit.
