International Journal of Scientific Research and Engineering Development

Traditional Business Intelligence (BI) platforms depend on human analysts to prepare datasets, detect anomalies, configure visualizations, and interpret results in business language. This manual dependency creates an insight-to-action gap that delays strategic decisions by days or weeks. This paper presents Agentic BI, an autonomous multi-agent framework that automates the complete analytics lifecycle through seven specialized cooperative agents orchestrated over a centralized state controller. Built on Python, FastAPI, and Google Gemini 1.5 Pro, the system performs schema detection, autonomous data cleaning, Z-score and IQR-based anomaly isolation, KPI computation, programmatic chart generation, predictive forecasting, and LLM-driven strategic narrative synthesis. Experimental evaluation on a 21 MB enterprise benchmark dataset comprising over 102,400 records demonstrates end-to-end processing in 9.8 seconds with 100 percent pass rate across thirteen test scenarios. Results indicate that agentic orchestration significantly reduces analytical latency while preserving data integrity, democratizing prescriptive business intelligence for non-technical stakeholders.

Akula Kavya Sri, Dr. P. Chiranjeevi, "Autonomous AI Agent for Business Intelligence: A Multi-Agent Orchestration Framework" International Journal of Scientific Research and Engineering Development, V9(4): Page(127-136) July - August 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.