AI inside the marketing agency: How CAPSBOLD is adopting AI

Most of the public discourse around AI in marketing still operates in extremes. On one end, the evangelists who promise that generative tools will replace entire departments. On the other, the sceptics who dismiss it as a glorified autocomplete. In practice, most agencies are finding a more balanced approach.

At the agency level, AI has quietly moved past the experimentation phase. Tools like ChatGPT, Midjourney, Perplexity, and a growing constellation of specialised platforms have become part of the daily toolkit – used not as novelties but as working instruments. The shift happened faster than most industry observers predicted, and it happened without much fanfare.

An internal survey recently conducted at CAPSBOLD, a creative marketing agency based in Limassol, offers a useful window into what this looks like in practice. The agency polled specialists across SMM, design, performance marketing, PR, brand strategy and creative production – and the responses paint a picture that is both encouraging and sobering.

The tool, not the talent

The most striking finding is attitudinal. Across every department and seniority level, the prevailing view of AI is deeply pragmatic. Nobody is starstruck. Head of Performance at CAPSBOLD, Sergey Tarasov, put it directly: “I don’t consider AI tools irreplaceable. They are simply a tool that, when used correctly, speeds up the time it takes to complete tasks.”

SMM Lead, Elizaveta Balakireva, echoed the point: “Nothing is irreplaceable – a new, better model will come along, and we’ll study it with interest.”

This matters because it signals maturity. When a team stops being impressed by AI and starts being precise about where it helps, the technology begins to deliver real value. The agency’s internal data bears this out: tasks such as research and media mapping have become around 35 per cent faster with tools like ChatGPT, Manus, Grammarly and Perplexity, while performance teams report spending roughly 13 per cent less time on routine work after integrating AI-supported automation.

In strategy work, AI has also begun to accelerate large-scale market research by helping teams gather and structure initial data layers significantly faster. On the performance side, Sergey Tarasov’s team now aggregates data from advertising dashboards, task trackers and internal chats using AI – freeing capacity that used to disappear into spreadsheets.

Creative production is changing fastest

Where AI’s impact feels most tangible is in visual and video production. A year ago, generating a coherent AI video – one where characters stayed consistent across frames – was a genuine technical challenge. Today, tools like NanoBanana have made it a viable production method. Within creative production, AI tools now allow teams to generate stable visual sequences and prototype video concepts far more quickly than before, enabling faster experimentation with characters, styles and formats.

This has practical implications for agencies and their clients. Projects that once required multi-day shoots or complex motion graphics pipelines can now be prototyped – and in some cases delivered – in a fraction of the time. But the risk is real: AI-generated content that looks generic, feels impersonal, or lacks the specificity that makes brand communication effective.

Several CAPSBOLD team members flagged this tension. Some team members note that audiences can often detect AI-generated content, which can reduce trust if it is used carelessly. For this reason, AI output is typically treated as raw material rather than a finished product, with creative intent and authorship remaining firmly human-led.

New skills for old jobs

One of the less-discussed consequences of agency-wide AI adoption is how it changes the skill profile of existing roles. The ability to write a clear, structured prompt – to communicate creative intent to a machine – is becoming a core competency. A vague brief produces vague results, whether you are briefing a junior designer or an AI tool.

Some specialists note that strong descriptive language – often developed through reading and writing – helps improve prompt writing and the ability to communicate creative intent to AI tools. This ability to describe ideas precisely has become increasingly valuable when working with AI systems.

In some cases, teams even use specialised AI assistants to generate and refine prompts, although human oversight remains essential to ensure the model interprets instructions correctly.

Meanwhile, entirely new roles are emerging. CAPSBOLD now employs an AI creator – a specialist who translates brainstorm concepts into precise prompts and delivers polished visual output. Oksana Baykova pointed to the broader rise of AI marketers, professionals who use AI to automate processes and sharpen campaign performance. These are not replacements for traditional roles – they are extensions of them.

The discipline behind the AI tools

Adopting AI is not simply a matter of subscribing to new software. It requires operational discipline – particularly around data privacy and quality control.

In practice, teams anonymise project data before it enters AI systems to ensure client information remains outside public training pipelines. A small procedural step, but it illustrates a broader principle: the faster you integrate AI into client work, the more carefully you need to think about governance.

Quality control is equally important. Specialists also report that AI tools frequently miss brand tone or occasionally produce inaccurate information, which makes human review essential.

The use of AI within the broader CAPSBOLD ecosystem extends beyond agency workflows. At sister tech company SPORTSOFT, AI tools are already supporting everyday processes across product and engineering teams. As Alexander Trushkin, Director of the Technical Department at SPORTSOFT, explains, project managers use AI to prepare project summaries, assess risks and track deadlines faster, while business analysts rely on it to structure requirements and transform discussions into clear user stories.

Developers use AI assistants to explain complex code, generate routine functions and troubleshoot issues more efficiently, while QA specialists apply them to draft test cases and analyse bug reports. “AI doesn’t replace engineers – it removes repetitive work and helps teams see the bigger picture faster,” he says.

What comes next

The marketing industry is still in the early stages of figuring out what AI integration actually looks like at scale. The tools will keep improving. But the competitive advantage will not belong to the agencies that adopt AI first. It will belong to the ones that build the sharpest judgement around when to use it, when to override it, and when to leave it out entirely.

As AI tools evolve, the competitive advantage will increasingly depend on how thoughtfully teams choose to use them.

About CAPSBOLD

CAPSBOLD is an award-winning creative marketing agency headquartered in Limassol, Cyprus, with offices in Yerevan, Armenia and Belgrade, Serbia. The agency delivers full-cycle marketing services – including brand strategy, creative production, performance marketing, SMM, PR, and AI-driven content – for clients across sports, technology, entertainment, real estate and beyond. CAPSBOLD has been recognised at the Web Excellence Awards, Symbol Creative Awards, and Cyprus Marketing Achievement Awards.