In partnership with
Yesterday, we talked about WPP's attempt to stay relevant by selling software instead of services. Today: a major brand showing why agencies might not even get that chance.
If you're feeling whiplash from how fast this is moving, you're not alone. You, me, all of us in marketing—we need to make some radical changes in how we think. I've got some thoughts on what that actually means at the end of this email.
But first, the news: Mondelez just announced it's invested over $40 million in a proprietary generative AI tool, built with Publicis and Accenture, that cuts ad production costs by 30% to 50%. They're already using it for social content for Chips Ahoy and Milka, and they're planning to deploy it for short TV spots by the 2026 holiday season and potentially the 2027 Super Bowl. By this November, Oreo will use it to generate product pages for Amazon and Walmart.
The message: if we're going to automate our marketing workflows, we might as well own the infrastructure ourselves. Have our Milka and eat it, too.
This is the logical endpoint of what WPP started with Open Pro. When agencies productize their expertise into self-serve platforms, they're essentially publishing the recipe. Smart brands look at that and ask: why pay for the meal when we can build our own kitchen? Mondelez isn't licensing someone else's tool or using an off-the-shelf platform—they're building proprietary capabilities that will compound over time. They'll own the models, the workflows, the learning. And critically, they'll own the cost savings.
The $40 million investment is substantial, but it's also probably less than Mondelez spends annually on agency fees for creative production across its portfolio.
They're not just cutting costs—they're shifting where the value accrues. Traditional agency relationships keep brands dependent on external expertise. Building your own AI infrastructure makes you less dependent, more agile, and able to iterate faster without negotiating scope or waiting for revisions. The tradeoff is you're now in the software business, which most CMOs didn't sign up for.
But if the alternative is watching your agency become a software company that charges you for access to tools they're selling to your competitors, maybe that's not such a bad deal.
Of course, Mondelez is being careful. No synthetic humans yet—they learned from Coca-Cola's disastrous 2024 AI Christmas ads. Everything still gets human review. They have guardrails against unhealthy messaging, manipulative language, and offensive stereotypes. But the trajectory is obvious: start with product page assets and social animations. Move to short-form video. Eventually, broadcast creative. The question isn't whether AI will replace significant portions of what agencies currently do. It's whether agencies or brands will own the tools that do the replacing.
So where does this leave the traditional agency model? Squeezed from both sides.
Holding companies like WPP are productizing their capabilities and selling direct-to-brand software. Brands like Mondelez are building their own AI infrastructure to bypass agencies entirely. The middle ground—high-touch service work that's too expensive to automate but not strategic enough to keep in-house—is disappearing fast.
And, look, I need to be direct with you about something.
You're going to have to start thinking like a software architect.
I know. I know that's not what you signed up for.
You got into marketing because you have taste, because you understand people, because you can tell a story. You didn't spend years learning how to nail a brand voice so you could spend your afternoons debugging data pipelines or writing model training specifications.
But here's what's happening: the people who do think that way are taking your budget. They're taking your headcount. And if you don't adjust, they're going to take your job.
Mondelez didn't spend $40 million building this infrastructure because they suddenly fell in love with technology. They did it because thinking in systems and architecture is now as important as having good taste. Maybe more important. Your edge isn't in crafting one perfect asset anymore. It's in building the machinery that generates a hundred variants a week, kills ninety-seven without drama, and iterates fast on the three that work.
That means understanding data architecture. Model tuning. QA automation. Measurement loops. Evals. Words no marketers had to know until now. It means shifting budgets from production costs to infrastructure. It means rewriting agency relationships around platform operations and IP training, not asset mills.
Your brain might not be wired this way. Most marketing brains aren't. But this is the final frontier now. You can learn to think in systems, or you can watch your CFO hand your budget to someone who already does. Those are the options. I wish there were more.
— Torsten & Peter
Field Note: The Hidden Skill: Thinking in Systems, Not Sentences
We were reviewing prompts from a client’s team last week — 40 variations of the same request, all written like casual DMs. No wonder the outputs were chaos.
Prompting isn’t about wording. It’s about architecture.
Few-shot examples, structured inputs, XML tags — these aren’t tricks; they’re how you encode clarity. Once you master that, you stop chatting with AI and start engineering behavior.
It’s the same shift every great leader makes: from giving tasks to building systems that think for themselves.
→ Inside Pro: Advanced Prompting Workshop — how to scale brand-consistent outputs across campaigns. Click here to upgrade.
The AI Insights Every Decision Maker Needs
You control budgets, manage pipelines, and make decisions, but you still have trouble keeping up with everything going on in AI. If that sounds like you, don’t worry, you’re not alone – and The Deep View is here to help.
This free, 5-minute-long daily newsletter covers everything you need to know about AI. The biggest developments, the most pressing issues, and how companies from Google and Meta to the hottest startups are using it to reshape their businesses… it’s all broken down for you each and every morning into easy-to-digest snippets.
If you want to up your AI knowledge and stay on the forefront of the industry, you can subscribe to The Deep View right here (it’s free!).
