Human+AI

Designing the marketing org your competitors are already building.

Jamie Skella · Skella & Co World Economic Forum Technology Pioneer MARK2026 · Kuala Lumpur linkedin.com/in/jamieskella · skella.com.au
Open warm, slow. "I'm not here to tell you AI is coming. You already use it. I'm here about the thing almost no one has done. Redesigning the organisation around it. Whoever does it first in your category doesn't get 10% better. They get a different kind of company." Land the title. This is about org design, not tools.

The uncomfortable part

Somewhere in your category, a competitor just shipped in a weekend what used to take you a quarter.

They didn't out-hire you. They out-designed you.

Springboard. Tell a short, concrete story. Use a real one if you have it, a small team you've seen out-ship a big one. The advantage wasn't budget or headcount. It was the shape of the team. Pause before clicking on.

Where everyone actually is

87%
use AI somewhere in the work.
8%
have redesigned anything around it.

Adoption is finished. Transformation hasn't started. That gap is the whole opportunity.

87% use GenAI in ≥1 workflow (2026, up from 51% in 2024); ~8% run autonomous multi-agent work. CMO Survey / Salesforce 2026.
★ HANDS-UP BEAT (do this before the numbers land). "Quick show of hands. Who's used AI in their marketing in the last month?" (most hands rise) "Keep them up. Now keep your hand up only if you've actually redesigned a workflow around it. Changed how the team works, not just added a tool." (most hands drop) "Look around. That's the eighty-seven and the eight, in this room." Then reveal the numbers. It makes the stat personal and gives the room a shared moment.

The spine of the talk. Everyone here is in the 87%. Almost no one is in the 8%. The 87% bought tools. The 8% changed the org. The rest of the talk is how you cross that gap, and why you have less time than you think. APAC adoption sits near 84%, so this room isn't behind on tools. It's on the start line for design.

The model we inherited

For fifty years, marketing scaled one way.
Add people.

Set the baseline everyone recognises. More reach meant more people, more agencies, more specialist layers. Capability was a function of headcount. Every org chart in this room is a monument to that equation. This is the assumption you are about to break.

What changed

Capability used to cost headcount.
It doesn't anymore.

The hinge of the talk. Say it slowly, let it sit. For the first time, a team can add capability without adding people. That single sentence breaks the operating logic underneath every marketing org chart ever drawn. Everything after this is consequences.

The unit of work is shrinking

What took a team of eight
now takes one.

Amazon / AWS agentic teams, 2026 (GeekWire). Projects that once took dozens of people now ship with a handful.
Concrete proof, brand names. This isn't a startup curiosity. It's Amazon, the company that invented the two-pizza team, now rebuilding it around agents. For marketing, the smallest unit that can carry an idea all the way to market is collapsing toward one capable human running a system. Don't celebrate or fear it. Just notice the floor is moving.

The mistake almost everyone is making

Bolting AI onto the org chart you already had.

Bolt-on

Same teams, same hand-offs, same approvals, now with a copilot in each seat. Faster typing. The system is unchanged.

Redesign

Rebuild the workflows, the governance, and who-decides-what around human + machine. The system itself gets smarter.

Name the mistake so the room sees itself. Most "AI transformation" is a faster horse, AI laid on top of a structure designed for humans only. You get slightly quicker execution and call it change. Real advantage comes from redesigning the operating system, not speeding up the old one. Regional proof they can't wave off as a US story is Grab. Anthony Tan is rebuilding the super-app as a "cyborganisation", putting AI agents straight into the hands of every merchant and driver while most companies are still piloting internally. Redesign, not bolt-on, happening in their own backyard.

Redesign, at super-app scale

Grab calls its new shape a

cyborganisation.

AI rebuilt into every function. An agent for every driver and merchant, while most companies still run pilots.

Anthony Tan, Grab, 2025 to 2026. "AI-first with heart."
The regional anchor. Say it with weight. This is the proof that redesign isn't a Silicon Valley story. Grab, the super-app this room uses every day, is rebuilding the whole company around AI. Anthony Tan calls it a cyborganisation. An agent for every driver and every merchant, in their hands, not in a pilot. Hold on the word and let them photograph it. Then make the scale point. A super-app does this with thousands of people. Keep that in mind, because later you show the same move done by one person, yours.

The reframe

AI isn't a tool your team uses.
It's a teammate. That changes what the team is.

The reframe. Co-intelligence. "AI for the team" keeps it a tool you point at work. The leaders pulling ahead treat it as a new kind of teammate and redraw the team itself. Mimi opened this morning on what only humans can do, empathy and listening. My structural question. How do you build an organisation that puts that human genius exactly where it counts and lets the machines carry everything else?

A new kind of headcount

Your next teammate
isn't a hire.

It onboards in minutes and works through the night. The job is no longer who to recruit. It is what to compose.

Make it tangible. Talk briefly about co-intelligent teams you've actually built, your own work. The CMO's job stops being a hiring plan and becomes a composition plan. What mix of humans and agents, pointed at what, with which human holding judgement. Keep it human. The people you already have matter more, not less, because they get to do the part only people can.

The ceiling just moved

The marketing department becomes a venture studio.

When execution stops costing headcount, the team is no longer a cost centre that runs campaigns. It becomes an engine for new brands and new revenue.

This is your blue-ocean idea. Nobody else on this agenda has it, so spend time here. The same capability that ships a campaign can stand up a whole brand. The most ambitious CMOs will stop counting campaigns and begin counting ventures launched. Marketing becomes the company's growth studio. This is the line that should get quoted back to you in the hallway.

Proof, not theory

My own brands are
powered by agents.

Crayon Dreaming
Generate a colouring page from any child's idea.
Snaffld
Automated inventory sourcing for collectibles.
Grab does this with thousands of people. Same move, minus the headcount.
The credibility move. You stop being a pundit and become a practitioner. Grab does it at super-app scale. You do it solo, across your own brands, where the marketing and growth are run by agents. Stay humble and concrete. The point isn't the size of these products. It's that the model already works at a scale of one. If anyone asks, say plainly what the agents handle day to day. Then turn to the hard question. Who holds the wheel.

Case study

Emu Money

A $200k marketing team,
run for $300 a month.

An Australian brokerage. The founder was about to hire a Head of Marketing and fund agencies. Now Emu Money uses a department of AI agents instead.

Emu Money, reported results over three months, 2026 · emumoney.com.au
The proof that lands hardest. Slow right down. Emu Money is a real Australian brokerage, 27 brokers. In March the plan was ordinary. Hire a Head of Marketing, hand them a big budget for SEO, paid and social agencies. Then the founder wrecked his knee, had surgery, and spent a few weeks with his leg up. In that window he built an LLM-agnostic team of agents and launched it as his new CMO. Three weeks of build. The human detail is optional, but rooms remember it. Then hit the contrast. A two hundred thousand dollar plan, running for about three hundred dollars a month. Let the number sit before you show what it produced.

What that department produced in three months

$170k
a month in new lead value
was 1 or 2 leads a day
11,960
organic visitors a day
was 80
660+
LLM citations a week
was none

It also cut its own cost per boosted engagement below $0.30. Was $0.80, tuned by the agents.

About 900 qualified leads a month at their $189 cost per lead. Citations across ChatGPT, Gemini, Perplexity, Copilot and Google AI Overviews. Emu Money, emumoney.com.au
The receipts. Read them plainly and let the numbers work. Organic visitors were 80 a day. Now close to twelve thousand. Qualified leads were one or two a day. Now more than thirty, about nine hundred a month, worth roughly a hundred and seventy thousand a month at their old cost per lead. And the fear sitting in this room, being invisible to AI. Emu had zero LLM citations. Now hundreds a week across every major model. It also publishes five to seven SEO articles a day and runs its own paid and social, all agent-run, with the founder approving. This one company is the whole talk.

Inside Emu's marketing department

Every seat is an AI agent. Claudia takes in the work and gates the quality. It all runs from a shared drive as readable files, fully human-auditable. Nothing reaches the founder until it clears the gate.

Claudia, a multi-agent marketing department. Emu Money, 2026 · emumoney.com.au
The how, for the sceptics. Keep it concrete. This is a real operating structure, not a black box. Claudia is the orchestrator, the head of marketing. Under her sit five specialist agents. SEO, content, paid media, a front-end developer for landing pages, and a reporting analyst. The whole thing runs from a shared Google Drive as plain files. Every draft and every decision is a readable file a human can open. The founder approves everything, and by the time it reaches him it has already passed the quality gate. It even learns, feedback shapes how the agents behave next time. The point for this room. This is designable. You can draw this org on a whiteboard. That is the job now.

The hard part nobody's solved

When agents can create
their own momentum,
who holds the wheel?

Deliberate gear change into the serious part. Everything so far is upside. Now the catch. Autonomy without design is chaos. Thin out the managers, hand work to agents and high-agency people, and the old control system disappears, the one where managers translated strategy down the line. So what coordinates the work now? Let it hang for a beat.

A flock has no manager

No standups. No roadmap.
No one in charge.
Perfect coordination.

Thousands of starlings move as one because every bird follows the same few simple rules. Your human and AI team is the same. No one can direct every person and every agent, so the strategy has to do the steering.

Make the analogy explicit, then land it fast. Point at the flock and say. "Watch a murmuration. Thousands of birds, no leader, no meetings, and they still turn as one. How? Every bird runs the same three simple rules, in real time." Then the bridge. "An AI-powered team works the same way. You can't put a manager over every person and every agent. So the coordination has to come from the strategy itself, simple enough that every human and every agent can act on it without asking." That is the setup for the next three slides. If the metaphor ever feels shaky, drop it and say the plain version out loud. When the team runs itself, the strategy is the manager. (Framing owed to Matt Ström-Awn.)

"Culture eats strategy for breakfast."

Unless you design a strategy
built to eat culture.

Reframe a cliché the room already believes. Culture usually wins because strategy is written to be filed, not used. But a strategy can be engineered to spread and to drive decisions on its own, and then it sets the culture. Two properties make that happen. Next slide.

How to build a strategy that runs without you

Salient enough to act on without permission.

Memetic enough to spread without enforcement.

Framework from "High-agency strategy" by Matt Ström-Awn, 2026.
The intellectual centrepiece, the framework you give the room to take home. Salience means the strategy guides a real decision today (clear and usable). Memeticity means it survives being passed around (frequent and accurate). High/high is the only quadrant that runs a high-agency, AI-enabled team, because there are no managers to translate or enforce. The other three all need management to work. Amazon's "two-pizza team" lives top-right. Most corporate strategy dies bottom-left. Your AI org needs top-right or it won't coordinate.

Designing the rules your system runs on

  1. 01Make it small enough to fit in one head.
  2. 02Ship the why with the what, so the rule can be rebuilt from first principles.
  3. 03Build in a test anyone can run. "Does this make us THE ___?"
  4. 04Repeat it until everyone says it without thinking.
Make it practical. These are the levers a CMO now pulls. The new craft is writing rules a human-and-machine team can run without you in the room. Small, so it fits in working memory. Reasoned, so it self-corrects when reality drifts. Testable, the way Southwest asks "does this make us THE low-fare airline?", a checksum anyone or any agent can apply. Repeated, so it's the default answer when no one knows what to do. This is governance for autonomy, and it's now core CMO work.

The job is changing under you

You no longer run
the campaigns. You
design the system
that runs them.

You become the architect of how a co-intelligent team works and improves, not its head of execution.

Bring it back to them personally. The biggest lever a marketing leader now pulls is not a campaign. It's the operating system the campaigns run on, the rules, the guardrails, the measurement, who holds judgement. Execution is more and more the machine's job. Design is more and more yours. The CMOs who get this stop being the bottleneck and become the multiplier.

Why this is urgent, not interesting

You have about 24 months before redesigned competitors make the old model uncompetitive.

Create urgency without hype. This isn't "AI is the future". A redesigned org compounds. It learns and improves on a faster clock, so a lead taken now is hard to close later. The window to be early is roughly the next two years. After that you're not gaining advantage, you're paying off debt. The cost of waiting isn't standing still. It's falling behind a curve that's bending up.

So, who runs marketing now?

The leader who designs
the system.

Jamie Skella · Skella & Co Strategy for agentic systems & emerging technology linkedin.com/in/jamieskella · skella.com.au
Close on the agenda's own question, answered. The room came in asking "CMOs vs AI-driven decision systems", as if it's a fight. It isn't. The winner is the leader who designs the system the decisions run inside. Build the org your competitors are still only talking about. Then open the Q&A, and if it's useful, "come find me, this is exactly the work we do at Skella & Co." Thank them warmly.
slides F fullscreen 1 / 23 Download PDF