Aston University and Phoebus Software Limited KTP 24_25 R4

A borrower or mortgage agent types a plain-English question about an account—"What are my remaining payments?"—and gets an instant, accurate answer from a private AI system, not a generic chatbot. This project addresses a specific problem in the financial sector: customer-service queries about loans, mortgages, and accounts are currently handled by humans or rigid automated systems that cannot interpret natural language flexibly. The result is long wait times, repetitive explanations, and frustrated users. The researchers are building an "Intelligent Service Platform" powered by a private Large Language Model—a type of generative AI that runs on the company’s own secure servers, not on public cloud systems. This means sensitive financial data never leaves the organisation’s control. If successful, the platform could let borrowers check balances, payment schedules, or eligibility in seconds, without navigating menus or waiting for a human agent. For the lender, it reduces call-centre workload and improves response consistency. The technology is not a general-purpose assistant; it is a narrowly trained tool for account-specific queries. The impact is on a quiet but essential part of modern life: the infrastructure that handles personal finance, where speed and data privacy both matter.