The Breakdown
Trend · AI agents move from making assets to running the plan
The robots aren't just making ads. They're buying them
AI: artificial intelligence
Amazon, Runway and Meta all launched AI (artificial intelligence) that plans, makes, publishes or optimises campaigns in the same week. What's left for marketers to do?
The TL;DR (too long; didn't read)
- Platforms now ask for a goal, a budget and creative, then decide the rest.
- Your edge moves to the inputs: strategy, creative and data.
- Human judgement still matters: brand, ethics, context.
01
What changed
Amazon's Full-Funnel Campaigns let its AI decide execution across formats. Runway's agent makes, publishes and learns from ads. Meta's Muse offers agent support to small businesses. The ‘buttons’ of media buying are being automated.
02
Where humans add value
Choosing the right goal, the audience insight, the creative idea, and knowing when the machine is optimising for the wrong thing (e.g. cheap clicks that never become customers).
03
The watch-outs
Black boxes: you may not see why budget went where it did. Platform bias: the platform's AI optimises inside its own walls. Always keep a way to test independently.
Read the original reporting
- How Amazon is streamlining ad buys across multiple formats with AI ↗
Marketing Dive · 29 Sept 2026
- Runway’s AI Video Agent Creates and Measures Ads Across Platforms ↗
Adweek · 30 Sept 2026
- Meta launches Muse AI for SMBs ↗
Social Media Today · 29 Sept 2026
- AI use among marketers hits 96% as focus turns to effectiveness, study says ↗
Marketing Week · 30 Sept 2026
The theory bit
The academic ideas that explain what's going on, with the full reference for your assignments.
Automation vs augmentation
AI can either replace human tasks (automation) or extend what people can do (augmentation). Augmentation tends to create more value and keeps humans central.
Here: Use AI to augment strategy and creativity, not just automate tasks away.
Brynjolfsson, E. (2022). The Turing Trap: The Promise & Peril of Human-Like Artificial Intelligence. Daedalus, 151(2).
Diffusion of innovations
New ideas spread through a population in stages: innovators, early adopters, early majority, late majority and laggards.
Here: With 96% of surveyed marketers using AI, we're past early adopters. The advantage now comes from how well it's used.
Rogers, E. M. (1962). Diffusion of Innovations. Free Press.
Over to you
No right answers here. These are to get you thinking. Jot your ideas down or chat them through.
- 1
What marketing decision would you never hand to AI?
Need a nudge?
Brand values, crisis response, ethics.
- 2
How would you know if an AI-run campaign was optimising for the wrong thing?
Need a nudge?
Clicks vs customers.
- 3
What skill will matter more for marketers in five years?
Need a nudge?
Judgement, creativity, data literacy.
Optional · about 10 minutes
Your mini task
In 60–120 words: design a fair test comparing an AI-automated campaign against a manually managed one. State the goal, how you keep it fair, and which metric decides the winner.