Семантический слой + AI-агенты новый уровень работы с данными (на базе DataForge)

During the webinar, we demonstrated how a modern data architecture is built using a semantic layer and AI agents. This isn't just a tool (more details at https://dataforg.ru/ ), but a holistic approach to managing corporate analytics—from formalizing business requirements to producing ready-made BI dashboards. The key idea of ​​the webinar is the transition from working with raw data and disparate reports to centralized business logic captured in a semantic layer. This eliminates discrepancies in metrics, ensures data trust, and dramatically accelerates the analytics development cycle. The demonstration detailed how DataForge implements a semantic layer through a register of metrics and dimensions (RMD). This is not just a catalog of metrics, but a formalized business model: unified metric definitions, calculation rules, relationships between entities, and logic control. This layer becomes the single source of truth for all downstream systems—BI, reporting, AI, and other analytical services. Special emphasis is placed on using AI to accelerate data processing. It demonstrates how AI agents automatically generate RPIs based on data structures and DDL, interpreting technical objects in a business context. This significantly reduces the entry barrier and the time required to create a semantic model, which previously required significant effort from analysts and developers. The recording also demonstrates how an AI agent functions as a data access interface. A business user can ask a question in natural language and receive a result in the form of an SQL query, a graph, or a ready-made dashboard. The AI ​​does not access tables directly, but rather through a semantic layer, ensuring the accuracy, reproducibility, and manageability of analytics. The solution's architectural framework is also discussed separately. At the center is the DataForge semantic layer, which integrates with DWH and BI systems. On top of this, an MCP server operates—a single point of secure AI access to data and business logic. This architecture allows for access control, standardization of AI interactions with data, and avoidance of chaotic, context-less text-to-SQL. An important takeaway from the webinar is that the semantic layer is becoming more than just a technical component, but a new corporate philosophy for data management. Without it, any AI scenarios (NLQ, RAG, agent-based systems) remain unstable and produce unreliable results. Only the presence of formalized business logic makes AI in analytics truly applicable to enterprise scenarios. The market context is also discussed: the growing role of agent-based AI and the transition to autonomous solutions. It is predicted that in the coming years, a significant portion of corporate systems will use AI agents for decision-making, making the issue of data manageability and semantics critically important today. Key layers discussed in the webinar 1. Data sources and DWH Operating systems, databases, and data warehouses are the foundation where source data is generated. 2. Semantic Layer (DataForge) A registry of metrics and dimensions, business logic, calculation rules, and unified metric definitions. 3. AI Data Access (MCP + Agents) A controlled layer for AI interaction with data via semantics, not raw tables. 4. Calculations and Analytics Generation Automatic SQL generation, metric construction, and the creation of data marts and aggregates. 5. Data Consumption (BI / Dashboards / API) Use of data in BI systems, reports, and user interfaces. 6. Self-service Layer for Business Business users gain access to analytics without developer intervention. Semantic layer, DataForge, AI agent, self-service BI, DWH, data warehouse, registry of metrics and measurements, MCP server, text to SQL, AI analytics, BI systems, dashboard building, data management, data governance, data modeling, semantic layer, NLQ, RAG, corporate analytics, data marts, analytics automation