A major study into how UK marketing agencies are adopting AI has revealed an industry racing ahead with hands-on experimentation but struggling to build the structures, skills and strategies needed to keep pace.
Prolific North collaborated with strategic AI builders twisted loop and research partners Mustard to produce The State of AI Adoption in UK Agencies, which draws on insight from 53 professionals across the UK’s creative and marketing sectors. More than half of respondents were founders, C-suite leaders or board members, with representation from junior and mid-level teams too. While responses came from across the country, most were based in the North, and the majority came from small to mid-sized independent agencies.
READ MORE: Revealed: Prolific North’s Top 50 PR Agencies 2025
The report provides one of the clearest benchmarks yet of how agencies are approaching the opportunities and risks of AI, as well as the clear gaps emerging between fast adopters and those still on the sidelines.
Findings from the survey showed an overwhelming 9 in 10 agency professionals now use AI, signalling that hands-on adoption has become mainstream. Yet, more worryingly, only 2% feel “very prepared”, painting a picture of a sector embracing AI but without the frameworks, governance or training to sustain it.
The tension is echoed across several findings. For example, 70% say efficiency is their top AI use case, but just 17% have built bespoke tools – pointing to low differentiation and a reliance on off-the-shelf platforms.
Likewise, 89% feel positive about AI, yet two-thirds fear over-reliance and the risk of creative sameness as everyone leans on the same models.
These patterns, twisted loop say, map almost exactly onto the classic innovation curve: 10% slow-to-adopt, 75–80% early adopters, and 10–15% building real capability.
Jonathan Summers, executive chair at twisted loop, said: “This data shows an industry leaning into AI with huge enthusiasm but most agencies haven’t yet built the scaffolding they need around that momentum.
“People are learning fast, experimenting fast, and asking AI to save time, but the structures, workflows and guardrails are lagging behind.
“What agencies should take from this is the need to shift from sporadic experimentation to deliberate capability-building. The differentiators now are AI fluency, workflow redesign and the ability to evaluate and direct AI. The agencies who treat this as an operating-model shift, not just a tooling exercise, will be the ones who gain a real advantage.”
Responses to the survey reinforce the theme of enthusiasm mixed with anxiety. Respondents described “approaching AI in an ad hoc manner,” and warned that the pace of adoption is “sending everyone into a frenzy” without stronger internal guidance.
While many talked positively about “making ops smoother and faster,” others raised concerns about losing creative personality or newer creatives becoming over-dependent on AI.
Skill gaps are emerging too with employees often outpacing leadership direction and pushing for clearer training, governance and defined workflows.
twisted loop’s analysis suggests this readiness gap could widen quickly. With big-tech AI capex projected to hit $571bn in 2026, compute expected to surge 1000x above GPT-3 levels within six months, and autonomous AI agents predicted to achieve eight-hour workflows by mid-2026, the underlying technology curve is accelerating faster than most organisations are prepared for.
Anthony Shephard-Williams, director at Mustard, said the dataset offers an unusually candid window into real-world agency behaviour. “What’s fascinating about this study is how consistently respondents expressed both excitement and caution,” he said. “You can see a sector genuinely energised by AI’s potential, but also very aware of the structural, creative and skills-based challenges that come with it.”
For Jonathan, the clarity of the patterns in the research was striking. “Agencies aren’t moving at random,” he said. “They’re clustering along a very recognisable innovation curve, and that gives us a strong foundation for tracking how maturity evolves from here.”
For agencies just getting started, twisted loop recommends investing first in accelerating foundational fluency: understanding how to direct, evaluate and decompose tasks for AI, alongside putting in place basic governance and building a simple list of possible AI use cases for the business.
For the majority “middle group,” the focus should be shifting from patching workflows with off-the-shelf tools to designing a clear AI blueprint and accelerating the development of even simple internal agents.
And for early leaders, the challenge becomes avoiding complacency by moving into deeper infrastructure: data fabric, RAG pipelines, agentic architectures and scalable AI operating systems.
As the capability curve accelerates, the report warns, uneven adoption risks becoming uneven competitiveness.
“twisted loop offers all of these AI value accelerator projects and more,” Jonathan said. “We’re AI builders who think, a team of exited tech founders that exist to ensure SMEs – not just large corporates – can obtain an AI advantage.”
Read the full report here.
