Sber explained why business needs an AI Overlay layer instead of restructuring departments

Sber proposes that companies should not break their organizational structure for AI, but instead add a separate technological layer—AI Overlay—on top of their existing business. The logic is simple: real returns come not from the number of pilots, but from an overall architecture that connects models, data, processes, and metrics into a unified system.

Why Pilots Are Stalling

The AI market has long moved beyond the experimentation stage. According to Roots Analysis, the global AI-agent market could grow from $9.8 billion in 2025 to nearly $221 billion by 2035 at an average annual growth rate of 36.

55%. Meanwhile, generative AI is already used by 88% of companies—in customer service, marketing, logistics, HR, finance, manufacturing, and analytics. Yet only 6% of organizations report stable growth in operating profit.

For business, this is an uncomfortable gap between technological hype and measurable financial effect. According to Sber, the reason lies in fragmented implementation. Typically, companies launch dozens of local initiatives: a chatbot in support, an assistant for marketing, document generation in back-office, knowledge base search for employees.

Each team solves its own problem separately, with its own data, integrations, and metrics. As a result, AI accelerates individual operations but does not reshape the entire end-to-end process. The effect disperses, costs grow, and it becomes increasingly difficult for management to understand where new profit actually appears and where there is simply more technological noise.

The Logic of AI Overlay

Instead of painful departmental restructuring, Sber proposes a different approach: lay a horizontal AI Overlay on top of the current structure. This is not a new ERP, not a replacement for CRM, and not an attempt to rewrite the entire business around a single model. It is about a layer that connects existing systems, data handling rules, and a set of AI services.

Such a layer allows agents and models to see the broader context, act according to unified standards, and participate not in isolated episodes but in the full chain of decision-making. The practical meaning is that a company can scale AI without organizational disruption. Functional teams retain their areas of responsibility but gain a common technological framework through which data, tasks, automation, and quality control flow.

If one scenario has proven effective in sales or support, it is easier to transfer it to adjacent processes. AI in this case ceases to be a set of disconnected pilots and becomes an operational layer of the business that can be developed, measured, and centrally controlled.

What the Layer Consists Of

For such an approach to work, it is not enough to buy access to a powerful model or implement a couple of bots. An architecture is needed where AI is embedded in the company's operating loop as a managed system. Sber identifies several mandatory blocks: unified access to data, an orchestration layer for models and agents, integrations with corporate platforms, security mechanisms, and clear result metrics. Only in such a configuration does AI influence not individual employee actions, but the speed, quality, and economics of the entire process.

• Unified access to data and knowledge
• Orchestration of models, agents, and scenarios
• Integrations with CRM, ERP, and internal services
• Security control, access roles, and answer quality
• Performance metrics and connections to P&L

If even one block is missing, the system quickly degrades. Without quality data, agents make mistakes and hallucinate; without orchestration, they duplicate actions; without integrations, they hit manual operation walls; and without metrics, the project turns into a pretty demo display. Therefore, implementing AI Overlay makes sense to begin not with abstract transformation, but with concrete workflows where you can calculate time savings, conversion growth, error reduction, or service acceleration improvement. That is where the horizontal layer best demonstrates its value.

What This Means

Corporate AI is shifting from hype over individual copilots to a platform approach. For executives, the question is no longer which bot to buy first, but how to assemble a unified AI framework on top of the existing business. Those who build such a layer earlier than others will more likely turn dozens of pilots into stable operational results without expensive company-wide restructuring.