A Day Without an Agency.

Why the communications industry will not fail because of AI, but because it is asking the wrong question.

On Tuesday 11 March 2031, Lena wakes up at 7.12 am. During the night, her personal agent has completed the weekly shop, switched her electricity provider and ordered shoes for her son. It rejected three brands. Lena will never know which ones.

She has not seen any advertising. Not because advertising has disappeared, but because it was filtered out before it reached her. Her agent decides on the basis of price, availability, provenance, returns policy and trust. It has no favourite brand. It has criteria.

At 8.30 am, Anna Berger, marketing director of a mid-sized drinks manufacturer, is sitting in front of her dashboard. 40,000 personal agents have assessed her brand. Sixty-two per cent recommended it. Thirty-eight per cent did not. No human being saw an advert. The communication was generated afresh for each individual contact.

Anna’s team consists of five people and a system. Its task today is not to brief a campaign. It is to decide which signals make the brand credible to families in southern Germany: provenance, regionality, packaging, responsibility. Anna terminated the last agency in 2028. Not because she was dissatisfied. Simply because nobody could say any longer what the agency was needed for.

At 9 am, Markus is sitting in his agency office. There is no brief on his desk. Not today, not this week. Markus did everything right. In 2025, he introduced an AI stack. In 2026, he reduced production time to one third. In 2027, he cut prices to remain competitive. His agency is more efficient than ever.

Only the marketing director who used to write the brief has become an operator. He runs a stack and buys capabilities on subscription. Efficiency was never the reason he called Markus.

Markus did not fail because of AI. He optimised the wrong question.

The disruption is not a scenario.

This story is invented. Its premises are not.

The GWA Spring Monitor 2026 reported that the member agencies’ gross income fell by an average of 2.7 per cent in 2025. Average profitability was 7.5 per cent. At the same time, the picture is not uniform: 43.7 per cent of agencies grew, while 50.7 per cent reported declines. The industry is not collapsing as one. It is polarising. GWA Spring Monitor 2026

The figures for insolvencies are even harsher. An analysis of insolvency announcements by iBusiness counted 198 agency insolvencies in Germany in 2025, compared with 89 in 2024. In the first quarter of 2026, the figure was reportedly more than 60 per cent higher than in the same quarter of the previous year. This is not a forecast about the industry’s future. It is a description of its present. iBusiness: Agency insolvencies in Germany remain high in 2026

And then there is Meta. In June 2025, Reuters reported, citing the Wall Street Journal, that Meta was working towards fully automating ad creation and targeting by the end of 2026. A product image or URL, a budget – and the platform generates the advertising materials, determines the audiences and optimises the delivery. On the same day, the share prices of several major agency groups fell: Omnicom by 3.2 per cent, Publicis by 3.8 per cent and WPP by 2.2 per cent. Reuters: Meta aims to fully automate advertising with AI by 2026

These figures can be interpreted in different ways. One can point to the industry’s resilience, to new fields of business and to growing demand for AI consultancy. One can say that technology has always changed work. All of that is true.

But it is not enough as an answer. This time, it is not just a tool that is being replaced. Three relationships on which the agency business is built are shifting at the same time: the relationship with production, the relationship with the client and the relationship with the audience.

Three attacks on a business model.

First: production is becoming a commodity.

Text, images, video, code and layouts can be generated in seconds. The problem is not that AI can do everything perfectly. The problem is that, for many tasks, it is quickly becoming good enough. The price does not fall only once quality is equal. It falls as soon as the quality is sufficient for the purpose at hand.

Demand for content will not disappear. On the contrary: there will be more content. But the price is shifting from creation to checking, selection and accountability. A campaign adaptation that can cost four or five figures today will become part of an ongoing approval and governance process tomorrow.

Second: the client is changing.

The traditional marketing director, who writes a brief and commissions services, will not simply disappear. But the role will change. Companies are building their own systems out of platforms, generative software, data and automated workflows. They no longer necessarily bring people into the organisation. They bring capabilities in – as software, as interfaces or as temporary expertise.

This is Inhousing 2.0. Not: we will now employ the same 30 people ourselves. Rather: five good people will operate a system that takes over a substantial part of the production previously carried out by an agency. An agency that offers execution alone is therefore no longer competing only with other agencies. It is competing with the client’s stack.

Third: the audience is being filtered by machines.

The campaign may still be produced and even paid for. Yet it may reach nobody. Personal assistants and shopping agents decide which offers enter a person’s field of attention at all. They assess structured information: price, availability, delivery time, provenance, trust and returns. This is not a complete description of future buying decisions. But it points in a clear direction: the recipients of communication are partly becoming systems themselves.

The agency is therefore losing not only access to the client. In the extreme, it is also losing access to the audience.

That is the tipping point. Each of these attacks could be absorbed on its own. Together, they strike at three points that cannot be adapted to one after another:

Point of attack	New competition	What comes under pressure
Production	Generative AI	Billable hours
Client access	Platforms and in-house systems	The intermediary role and the brief
Audience access	Personal agents and answer engines	The audience for the campaign

Agencies are currently optimising the rudder of a ship whose river is drying up.

What becomes cheap first.

The most exposed services are those that are standardisable, data-driven and deliverable without deep client context: content production, adaptations, localisation, standard web development, reporting and parts of the performance business.

Not because these services are worthless. But because their value has often been sold through the amount of work involved. When the machine drastically reduces the amount of work, that pricing model loses its foundation.

The classic agency often responds with the same answer: more new business. It is understandable. The GWA Spring Monitor identifies winning new clients as the most important lever for positive business development. But new business built around the same offering is not transformation. It merely spreads the pressure across more pitches.

The uncomfortable question is therefore not: How can we produce faster with AI? It is: What remains when production is no longer the reason anyone calls?

The answer is not automatically creativity. That would be too convenient. The industry has been saying for years that creativity will remain human. That may be true, but it is not a business model.

Other things will become more valuable: context, judgement, conviction, accountability and the ability to implement decisions within a system.

Everything given away today will become valuable.

The real turn is this: Everything agencies bill for today will become cheap. Everything they give away today will become valuable.

Strategy becomes valuable when it is no longer an introduction to production, but a robust decision-making system. Brand becomes valuable when it is not merely a set of assets, but a source of credible signals for people and machines. Creativity becomes valuable when it is understood not as the production of variants, but as a rare, accountable act of setting a direction.

This leads to three consequences.

1. Identity: from self-description to purpose.

The positioning statement ‘We are creative, digital, strategic and approachable’ is weak not simply because it sounds dull. It is weak because it says nothing about the world outside the agency.

A viable positioning will answer a different question in future: what problem becomes more likely without us?

Not: ‘We create brand communications.’ But: ‘We make family-owned businesses trustworthy in the agent economy.’ Not: ‘We offer full service.’ But: ‘We build brands that cannot be generated at will.’

This is not a linguistic exercise. It forces a decision about domain, conviction and accountability. Agencies will no longer be judged primarily by what they can do. They will be judged by what they stand for and in which problem they are unmistakably useful.

2. Structure: from people-based operation to systems operation.

The agency of the future will not necessarily be larger. It will be larger in a different way.

A small senior core will take responsibility for judgement, client relationships and domain expertise. An in-house production system will handle repeatable work. Partner networks will replace the full-service model. Methods, brand systems, data models and tools will become assets – not applications installed somewhere on employees’ computers.

This could be an agency of five to eight people enabling the output of a former 30-person organisation. The number is not the point. The point is whether the agency owns a system that can survive beyond individual people.

That raises a question the industry has so far avoided: who will train the senior people of 2035 when junior work has been taken over by machines?

A future-proof agency therefore needs a training model that does not merely teach output. From an early stage, junior people must learn to justify decisions, tolerate contradictions, understand context and take responsibility. Whoever organises this is not just developing staff. They are building judgement as a competitive advantage.

3. Growth: from projects to accountability.

The revenue model needs to turn: production becomes an add-on, judgement becomes the product.

In practical terms, this might mean standardised brand governance audits, agent-readiness checks, trust programmes, brand management systems and retainers for ongoing direction. What is paid for is no longer the number of hours, but the quality of decisions and accountability for their consequences.

This can be complemented by revenue shares, success-based models and proprietary intellectual property. Not every business needs to become a principal model. But an agency that exchanges time exclusively for money becomes vulnerable the moment time is no longer a valid measure of value.

The agency does not necessarily have to become a management consultancy. Its opportunity lies precisely in remaining closer to the brand: at the point where conviction, organisation and market promise come together.

Four forms of agency in 2031.

Four models are emerging from this shift.

The trust architect builds brands that people and systems read as credible: through provenance, conviction, data quality and reputation.

The systems partner designs and operates the client’s communications and production system together with them – including rules, technology, governance and training.

The creative atelier works deliberately small, human and expensive. It does not sell variants, but signature: the unexpected, the cultural, the things that do not emerge from an optimisation loop.

The transformation partner does not sell communication as an isolated measure. It helps organisations use brand as a system of leadership and decision-making.

None of these forms primarily sells production. All of them sell judgement, accountability and context. And none needs to wait until 2031. They already exist in outline today.

The question that matters now

Markus would not have needed to know exactly what the world would look like in 2031. He would not have needed to anticipate every new technology. He would only have needed to recognise that efficiency is not a purpose.

A more efficient agency remains an agency if it merely does the same thing faster. The real task is to redefine its value before the market does it for them.

The question is therefore not whether an agency uses AI. That will soon cease to be a strategic distinction. The question is whether, in 2031, anyone will have a reason to call it.

Looking inward is not a strategy. The view from inside shows what an agency can do. Only the view from outside shows what it is worth.

_________________________________________________________________

Sources and notes

• GWA, Spring Monitor 2026: revenue development, profitability, polarisation and transformational pressure: https://www.gwa.de/presse-meldungen/gwa-fruehjahrsmonitor-2026/
• iBusiness, Agency insolvencies in Germany remain high in 2026: analysis of insolvency announcements: https://www.ibusiness.de/branche/db/817861jg.html
• Reuters, Meta aims to fully automate advertising with AI by 2026, WSJ reports: report on Meta’s automation plans and the reaction of agency share prices: https://www.reuters.com/business/media-telecom/meta-aims-fully-automate-advertising-with-ai-by-2026-wsj-reports-2025-06-02/

The scenes set in 2031 are a journalistic piece of speculative storytelling. The people, figures and processes described are neither a forecast nor an empirical case study. They translate observable technological and economic developments into a deliberately pointed everyday situation.