Accessing and analyzing data is often a bottleneck for non-technical users, requiring technical skills and time-consuming processes. A multi-layered AI agent architecture, powered by Natural Language Processing (NLP), addresses these challenges by enabling conversational interactions with data. This architecture consists of three key components: the Interface Agent, which interprets user queries and structures them into actionable blueprints, a State Manager that managers the query state, and Query Agents, which execute analyses through systematic reasoning and iterative sub-queries. This multi-layer architecture ensures logical, transparent, and efficient data analyses.