AIBDO-X: AI Powered Business Decision Optimizer Executive Tool

Abstract

The traditional business intelligence (BI) tools and products, like Tableau, Microsoft Power BI, Qlik Sense, Google Looker Studio, SAP BusinessObjects or IBM Cognos, are very much geared towards descriptive charting and historical KPI tracking. These tools indicate the events that occurred and when they occurred but they do not provide the "why" or the suggestions for management action that could correct a metric's decline. Root cause identification, inter-departmental bottlenecks identification, and planning corrective interventions remain completely manual. In real life situations, enterprise teams also deal with separated workflows where they jump between separate data processing, predictive modelling, time-series forecasting and executive reporting solution software. To close this, we are introducing the AI Powered Business Decision Optimizer Executive Tool (AIBDO-X). This zero cost Decision Intelligence platform, developed with a FastAPI back-end, orchestrates 22 special purpose AI agents via over 30 modular Python files. The sites include user uploaded raw CSV or Excel files from 7 functional areas: Marketing, Finance, Sales, Human Resources (HR), Supply Chain, Customer Operations, and Risk. At Anasys, AIBDO-X implements a ten-step analytical workflow for each department running in nearly 50 seconds. This automated pipeline performs schema validation, missing data imputation, exploratory data analysis (EDA), domain-specific feature engineering, and competitive benchmarking over over a dozen machine learning estimators and multi-horizon forecasting with models such as ARIMA, Prophet and Holt-Winters. The system does not generate black box predictions, but instead it combines the capabilities of a Root Cause Agent (Root Cause Analysis, feature importance using SHAP, LIME and statistical correlation analysis) and an Explainability Agent that transforms the understandability of feature attributions into plain language business insights. Seven domain-specific Strategic Agents then evaluate whether each of these departments' operations is feasible, then a central Decision Agent integrates them all into one executive action plan. All past choices are recorded in a persistent Memory Agent to track events in the institution. Last but not least, a fault-tolerant LLM Manager ensures requests are sent to multiple API keys for Google's Gemini 3.5 Flash and Groq GPT-OSS-120B APIs - even if rate limits, it proves to be deterministic LLM rules act which also enables uninterrupted, zero-cost operations.