The starting point
Business challenge
Data access and interpretation were separated: users had to locate figures, reconcile sources, prepare reports, and then investigate drivers of changes. A generic AI assistant could not safely be treated as a source of financial truth.
Alexander’s role
Work and solution
Alexander developed a controlled reporting architecture in which AI uses authorised tools and Model Context Protocol connections to retrieve current facts from corporate sources. He defined flows for querying Odoo, PostgreSQL and SQL datasets, CRM, accounting, sales, inventory, spreadsheets, and external data; combining financial and operational indicators; analysing variances and trends; and producing management explanations in natural language. The design includes access restrictions, validation, approval, auditability, and human responsibility for critical conclusions.
What changed
Project result
The resulting architecture enables conversational analysis to be grounded in authorised business data rather than values embedded in prompts or static text. Managers can investigate questions such as margin changes, plan-versus-actual variance, customer performance, pipeline conversion, and cost movements through a controlled analytical layer.
Business effect
What the work made possible
Reduces dependence on manual data gathering, shortens the path from a management question to an evidence-based explanation, improves consistency between operational and financial analysis, and supports repeatable reporting with appropriate governance.