Summary: "AI or knockout" is the headline of the current agency debate. The truth sits in between - AI replaces execution, not judgment, and the real dividing line runs not between agencies with and without AI, but between agencies with and without systems.
Right now the industry is debating this in big words: "AI or knockout," one commentary headlines, calling 2026 "the year of truth" for agencies. Another piece claims the opposite of what you'd expect: AI won't replace influencers, but it will replace agencies. A third contradicts both and says the real problem isn't AI at all, but that many agencies are standing in their own way. Three readings, one question. I have an unusual vantage point on this: I co-founded and ran Kampart, a classic communications agency, until it was wound down this year – and I run Luftschloss, a studio built from day one with AI as a core tool. I know both sides of this debate from the inside.
What's right about the alarm
The part of the debate that holds up: many tasks agencies have sold as expertise really are becoming basic technology. Research, first drafts, reporting, monitoring, deriving a strategy from raw data – AI can now do all of that in a fraction of the time a junior team used to need. Whoever built their business model on exactly these tasks is feeling that pain right now. This isn't a future scenario, it has already happened – to us too.
What's overblown about the alarm
The conclusion "therefore the agency becomes obsolete" strikes me as wrong, for the same reason the influencer marketing debate is currently making clear: people follow people, not avatars. Trust is built through relationship, not automation. That's as true for influencers as it is for agencies and their clients. A managing director doesn't hire an agency only for execution, but because they trust a judgment – on visual language, tone, the right moment for a campaign. That's exactly what I experienced in practice: defining visual language as a system at Kampart was never an execution task, it was a matter of judgment – and that doesn't disappear just because execution gets faster.
The real dividing line
The most convincing argument from the current debate: the real competition isn't between agencies and AI, but between agencies with systems and agencies without systems. That matches my own experience exactly. Many agencies don't fail in the market, they fail because of themselves – lack of process discipline, projects hanging on individual people, gut decisions instead of metrics. AI only sharpens this problem, it doesn't create it. A team without systems gets faster and more chaotic with AI, not more orderly.
For us, that's exactly why we invested early in our own internal tools – from our central tool index to our acquisition CRM. Not because software alone saves a business model, but because systemization is the precondition for AI to have any leverage at all. Whoever doesn't know their processes can't speed them up either.
What remains when execution gets cheap
When research, reporting, and first drafts become commodities, value shifts to where it actually always sat, even if it was long hidden under execution costs: network, curation, accountability for outcomes, and the ability to recognize the one right path for a specific brand among many possible ones. That's the same shift I described for custom software: when implementation becomes cheap, the knowledge of what should even be implemented becomes the scarce resource. For agencies that means: whoever only sold execution gets replaced. Whoever sold judgment, curation, and accountability becomes more in demand, because they can now deliver faster.
What I take from this for Luftschloss and Gieske Studios
The practical consequence isn't a doom scenario, it's cleanup work: first look honestly at your own processes, then use AI where it actually saves time without taking over the decisions that genuinely need judgment. We've since started teaching this exact sequence externally too, in our AI seminars for companies – because the same distinction that decides whether an agency survives also decides whether an AI project changes something inside a company or just produces arbitrariness faster.
Conclusion
"AI or knockout" is the wrong question, because it assumes every agency faces the same choice. In reality, AI hits agencies very differently – depending on whether their business model rested on execution or on judgment, and whether they work with systems or with improvisation under constant stress. Whoever has neither will genuinely have a hard 2026. Whoever has both will find in AI exactly what the name promises: a tool, not a replacement.