Marketing teams at Russian companies are increasingly shifting to digital tools.
Artificial intelligence has evolved from a simple tool into a systemic participant in the creative process, taking on routine tasks and amplifying a team’s creative potential, Konstantin Medvedev told IT RUSSIA. Medvedev heads product and content promotion at SberMarketing.
“Any piece of content starts with an idea – without it, there’s no text, no visuals, no discussion. Instead of traditional brainstorming, we use an ‘intellectual sprint’ approach powered by neural networks like GigaChat and Perplexity. The value of AI at this stage isn’t in ready-made solutions, which are rarely successful, but in its ability to spark a specialist’s own idea. AI also acts as an intelligent assistant: it gathers information from multiple sources, builds trend lists, and summarizes large volumes of material, significantly speeding up the pre-production phase,” Medvedev said.
According to him, an idea becomes truly valuable only after it is tested and refined so that it speaks the brand’s language and solves its specific tasks. This is where principles of controlled AI use come into play.
“To make sure all texts sound like they’re truly ours, we trained the neural network on the brand’s most successful publications. Now, when AI produces a draft, we can apply this ‘filter’ to bring it into the right shape. Still, the final call on whether the text hits the mark in terms of meaning and values is always made by a human. That said, we clearly see that AI significantly reduces the time needed to prepare a single post,” Medvedev noted.
New Design and Smart Metrics
The rise of AI has also changed how design works in marketing. Designers are no longer artists drawing everything by hand; instead, they increasingly resemble technologists who configure the process of creating visuals. The goal of design, Medvedev said, is not to replace meaning but to amplify its impact.
“To identify strong ideas, we combine two approaches: manual analysis of metrics and the use of neural networks to detect patterns and insights. This way, we understand not just whether a post ‘worked,’ but why. All posts are evaluated regardless of how extensively AI was used in creating them. We rely on the analytical capabilities of the neural networks themselves. For example, advanced modes in Perplexity and GigaChat help us dive into large datasets from TGStat and Medialogia. The goal isn’t just to collect statistics, but to uncover hidden connections and patterns – what exactly in a topic, format, or timing makes an audience respond. This allows us to systematically optimize our content strategy,” Medvedev emphasized.
He added that the ideal marketing model to aim for is a fully integrated cycle, where raw material generated by an AI agent is automatically turned into a publication-ready layout. In that setup, the human role is reduced to a final strategic review and oversight, ensuring nothing important is missed.
