An AI Agent-Powered Gift Shop
sellers of flowers, sweets and gifts operating on the Russian marketplace Flowwow have gained a new tool for managing their online stores through the Seller mobile app.

Store owners can manage their shops through a text-based chat with an AI agent. Before taking any action, the AI must ask the user for confirmation. The solution is built on the open MCP protocol, which connects the agent to the platform's API.
Core Features
The MVP includes 13 features: creating and editing product listings, changing prices, updating inventory and retrieving product information. To get started, sellers simply add Flowwow's MCP server to the AI system they use and complete the authorization process.
“The tool's main goal is to make sellers' day-to-day work easier. We want them to spend less time navigating their seller dashboard or learning how to work with APIs. Instead, they can simply type a request into an AI chat, such as ‘show me all the rose bouquets’ or ‘update the photo.’ Delegating these tasks lets business owners focus on what matters most: growing and improving their businesses,” said Roman Albach, product manager for partners at Flowwow. Smaller shops stand to benefit most in practical terms, since their owners often manage inventory, pricing and product assortments themselves.

Agentic Commerce
In August of this year, DataInsight and Yandex experts examined the emerging landscape of agentic commerce in the Russian market. Their study concluded that by 2028–2029, sales involving AI agents could reach 400 billion to 850 billion rubles (approximately $4.9 billion to $10.3 billion), or 2–4% of the e-commerce market. Over the following two to three years, the figure is projected to rise to 2–3 trillion rubles (approximately $24 billion to $36 billion), equivalent to 7–11% of the market. In other words, AI is gradually becoming a sales channel.
Four models are likely to coexist and compete for consumers in the online commerce market: generative AI, ecosystem AI, marketplace AI and AI designed for retailers. In one form or another, Russian companies are already putting all four into practice.
Meanwhile, AI assistants for sellers will continue to evolve, helping businesses organize their operations more efficiently and spend less time on routine tasks. More and more of that work will be delegated to artificial intelligence. The spread of MCP will make it possible to connect AI agents to CRM and ERP platforms, warehouse management systems and order management tools. An agent, for example, could reconcile warehouse inventory with marketplace records and suggest catalog updates. This model of e-commerce development will require Russian IT companies to build new tools for integration and security.

From Listings to Operations
The technological foundation for AI agents capable of interacting with external software was laid in 2024 with the creation of the MCP protocol. It allows language models to integrate with enterprise systems, databases and applications.
In 2025, leading AI developers adopted MCP as a standard, paving the way for broader deployment. That same year, VkusVill, a chain of brick-and-mortar and online stores, launched an experimental MCP server that gives AI access to its product catalog. This made it easier for customers to find the products they need and check their specifications. In 2026, the Ozon marketplace introduced a “Smart Assistant” in its seller dashboard to answer customer questions and recommend products. Online retailer VseInstrumenti.ru opened its catalog to AI agents, allowing them to select products in response to customer requests. Within the Yandex and Sber ecosystems, AI is already doing more than helping users choose a product or service: it is guiding them through to a purchase.

Automation With Human Approval
Marketplaces are using AI tools to compete not only for customers but also for sellers, as the Flowwow case demonstrates. Entrepreneurs stand to benefit: automation saves time and allows them to focus on strategic priorities while leaving routine work to algorithms. Growing businesses will not need to hire additional employees to handle that workload.
Over the next three to four years, MCP, APIs and other tools that allow AI agents to interact with commerce platforms are likely to see increasingly widespread adoption, while the agents themselves will gain significantly broader capabilities. They will connect with enterprise resource planning and customer relationship management systems, as well as accounting and banking software. Entrepreneurs will be able to perform a range of operations through a single interface simply by issuing commands to an AI agent. Even so, decision-making, strategic customer relationship management and oversight of AI-driven sales will remain in human hands.









































