The Story Independent Agencies Should Study
One agency got more attention than any other in Spark AI's 2026 report, and it's not a network shop. It's a production agency called Tuncarp.
Here's what they did. They restructured their entire delivery model around AI-enabled roles — strategy, operations, creative, engineering. They rebuilt high-end video work that previously cost around £20,000 per asset in traditional production, and started delivering it for around £5,000 per asset.
They kept margins of 70 to 80%.
Publicis subsequently appointed them as its dedicated AI content studio — meaning one of the biggest networks in the world hired a small independent as the specialist supplier for its AI-enabled content work.
The quote from MD and founder Chris Murphy, captured in the Spark report, explains the culture behind it: "You need people who are willing to go down rabbit holes. Curiosity is the real skill now."
That's the whole story. Let me explain why it matters if you run a creative team.
The Hidden Mechanic: Pricing by Asset, Not Hour
Most agencies reading Tuncarp's numbers zero in on the cost reduction. £20K to £5K. 75% reduction. Sounds like commoditisation.
It's not. It's the opposite.
The mechanic that holds Tuncarp's margin at 70 to 80% isn't the cost saving. It's the decision to price per asset, rather than by the hour. Under an hourly model, if AI made the work faster, the client would pay less. Period. The margin would compress toward zero.
Under a per-asset model, when AI makes the work faster, the agency captures the efficiency as margin. The client still pays for the outcome. The agency still hits the deadline. The difference is the agency makes more money, not less.
Pricing structure is what determines who benefits from AI efficiency — you or your client. Hourly-priced AI work is a race to the bottom. Output-priced AI work is a margin expansion play.
What Changed Operationally
Restructuring around AI-enabled roles is the part that scares most agency leaders, because it sounds like headcount reduction. Read it carefully. It isn't.
Tuncarp didn't fire its production team. It redefined what production work means.
Under the old model, producing a £20K video required producers, designers, 3D artists, editors, and multiple rounds of external partner coordination. Under the new model, AI handles significant portions of concepting, iteration, versioning, and review-prep work. The human roles shifted upstream — more strategy, more direction, more quality judgement, less manual execution.
The result is a leaner team shipping the same output, at a fraction of the old cost, with higher margin per person. The team didn't shrink — its role expanded.
That shift from execution to direction is the same pattern we see in every successful NotContent transformation. The team that used to execute what a director briefed is now the team that directs AI, curates output, and ships. Their craft doesn't disappear. It gets promoted.
Why Publicis Hired Them
The second thing worth studying is the network-to-independent supplier relationship Tuncarp won.
Under the old creative industry logic, a big network with internal production capability shouldn't need to hire a 20-person independent as its content studio. The network has scale. The network has buying power. The network can build it in-house.
That logic has broken.
The Spark report is explicit about why. Networks are investing hundreds of millions in AI infrastructure that has to serve thousands of people across dozens of markets. They can build platforms. They struggle to build focused specialist capability that runs fast. An independent agency with 30 people and one distinctive operating model can compound expertise tighter than a global network can.
That's what Publicis bought when they hired Tuncarp. Not just delivery. Specialist focus that the network's own structure can't reproduce.
If you're an independent creative team, this is the strategic opening. You don't need to become another agency trying to out-bloat the networks. You need to become a specialist so tightly positioned, with such a clean AI-enabled operating model, that a network would rather hire you as a supplier than try to build your capability internally.
What This Doesn't Tell You
Let's be clear about the limits of the Tuncarp story.
Their margins are 70 to 80% on a specific kind of work — production-focused, asset-based, clearly scoped. If your agency sells ongoing strategic partnership, retained creative direction, or campaign-level thinking, you don't just switch to per-asset pricing and keep the same economics. The pricing model has to match the kind of value you're actually selling.
Spark's four archetypes frame this well. The Creative Differentiator sells thinking. The Operational Realist sells efficient delivery. The Systems Architect sells IP and infrastructure. The Performance Engine sells scale. Each needs a different pricing approach, and only some benefit from the Tuncarp playbook directly.
But every agency can take something from this story, regardless of archetype.
The Three Questions This Quarter
For any creative team trying to figure out what Tuncarp's example means for them, three questions are worth answering this quarter.
Which parts of our delivery are actually asset-based? Not everything. But something is. Find it, and price it like Tuncarp.
Where would AI compression threaten our margin under the current pricing model? That's the first conversion. Get ahead of it.
What is the specialist positioning that would make a network hire us rather than build us? That's the strategic horizon. If you can't answer it, you're competing on price.
The £20K-to-£5K story isn't about doing the work cheaper. It's about building the pricing structure that captures AI efficiency as margin instead of giving it away.
Decide whose side of the ledger that efficiency is going to show up on.