AI starts with data, not a model

I spoke at the TAdviser conference about how we are building the holding's IT landscape at Roskhim. The main idea: all systems are gathered around a single core. The core is not a "fashionable platform" but three boring things: an MDM system, a data bus and a corporate data warehouse. They provide a reliable integration foundation on top of which AI and RPA are layered. Everyone argues about which model to choose, but in industry AI is limited not by the model — but by data and its context. That is why we are working on an "enterprise ontology": a structured description of the business. Without it, autonomous agents will produce beautiful answers to the wrong questions. The second layer is people: moving from coding to analytics and data management. The third is personal: immersing myself in technology (AI agents, Deep Search) and networking. A leader's job is to turn innovation into roadmaps of change.