AI Creative Production Goes Mainstream for Mid-Size Marketers

For years, generative AI creative production looked like a superpower reserved for the biggest budgets. That assumption is breaking. Wyndham Hotels & Resorts, consumer healthcare company Opella and betting brand BetMGM are each building in-house creative teams that pair AI-generated assets with media performance data—and they are doing it at speed.

The pattern matters because it moves AI creative beyond the first-mover set. Early adopters such as Unilever, Coca-Cola and L’Oréal had industry-leading budgets. The newest wave has smaller, but still meaningful, media muscle.

The upper middle finds its AI lane

Wyndham’s overall marketing and loyalty budget was $565 million last year—hardly small, but far below the multi-billion-dollar budgets of AI first movers. L’Oréal, for one, spent $1.98 billion on AI investments alone in the same period. Yet Wyndham’s five-person in-house team now uses AI firm Adora to generate paid and owned Instagram assets, from video to still images.

That team reports producing 15 times the volume of assets it did before, while cutting concepting, production and approval time by 75%. Wyndham has also seen a six-times improvement in site visit rate over its benchmark as it increased Instagram media spend.

Opella, spun out of Sanofi in 2025, runs an 85-strong team using generative AI for strategy briefs, asset generation and animation. It now produces roughly 20 times the creative it did before—more than 20,000 pieces of content—including AI versions of the allergy brand Xyzal’s owl mascot Nigel.

BetMGM takes a more cautious route. With around $160 million in annual media spend and 120 in-house creatives, it uses tools including Arcana AI for imagery and video, but keeps some production human in its tightly regulated category. The brand will not let AI generate assets carrying betting odds, for now.

Why the middle can move faster

Brandtech Group’s Pencil platform is one backbone for these systems. Its fastest-growing client segment is advertisers with less than $1 billion in annual income. Pencil CEO Will Hanschell puts it simply: “They culturally move faster.” These brands want something bespoke but less rigid than a traditional enterprise solution.

Costs are also falling. The price American businesses pay per million tokens has dropped 41% from its March peak this year, making the technical lift easier to outsource. Generative tools are also getting better at visual consistency, which lowers the barrier for brands worried about off-brand output.

• Wyndham: 15x more assets, 75% less production time, 6x site visit rate.
• Opella: 20x creative volume, 20,000+ pieces of content.
• BetMGM: $160m media spend, humans retain regulated creative decisions.
• Pressure: 81% of CMOs expect to need significantly more content; Kantar’s early-stage creative tests rose 17% between December 2025 and July.

Volume is the easy part; trust is the moat

All three brands impose guardrails. Opella has rules around depicting medical professionals and adds AI disclosures in markets such as the EU. “No fake doctors,” says Julien Leterrier, global head of martech at Opella. BetMGM’s senior director of creative, Victor Gonzalez, sums up the restraint: “You can’t hold AI accountable.” Wyndham is still debating whether AI-generated people belong in scenes of its hotels and resorts.

That tension is the real lesson for brand teams. AI makes producing variants cheap. The strategic edge comes from deciding what should be automated, what should be tested before media dollars are spent, and what still needs a human sign-off.

The next wave is likely to pair assembly-line production with more early-stage testing. Kantar’s 17% jump in early-stage creative tests suggests marketers are using AI to produce more, then filtering before budget is committed. CMOs who treat AI output as hypotheses rather than finished ads will get the volume without eroding trust in regulated or sensitive categories.

Source: Digiday