The Future of Credit Decisioning in the Agentic AI Age

By Muhammad Hejvani, CDTO at MOGOPLUS

While the burning heat around Artificial Intelligence (AI) and Generative AI have started to cool a bit, there’s indisputable evidence of its potential to transform credit risk and decisioning. As we enter the ‘disillusionment’ phase of the AI ‘hype cycle’, according to a recent Gartner report, we observe solid examples of how AI can enhance both quality and efficiency of the credit decisioning process. A McKinsey & Company study highlights how AI can be leveraged in data preparation and analysis, noting its ability to flag anomalies, improve data quality, and summarise key metrics for credit officers. They also expect that AI agents to be able to perform a series of tasks autonomously, from data ingestion to the analytical operations required to create credit memos with minimal human involvement. Capgemini has also reported a significant increase in automated decisioning by 70-90%, an overall increase in approval rates by 15-40%, while lowering the loss rate by 10-25% for the companies which have adopted the AI-powered solutions.

From a practical standpoint, there are still questions on how to effectively use AI to support the credit decisioning while addressing legal, regulatory, and compliance issues. Although numerous AI-powered capabilities and large language models (LLMs) have been introduced to the market in a short timeframe, we haven’t yet seen an adequately robust reasoning engine that can fully automate the decision-making process. What’s more, even with capable reasoning and cognitive engines, we must address key compliance issues such as explainability, impartiality, and other forms of potential biases, before fully utilising them in AI-powered credit decisioning support systems.

To explore the role of AI in credit decisioning in more depth, I believe we must explore the following topics a bit further:

• The role of automation and decision support systems in the credit and loan application process.
• The current and future state of data analytics and digital transformation in credit decisioning, focusing on cloud and AI solutions.
• Recent advancements in analytics and AI, specifically Generative AI and Agentic solutions, for credit decisioning.
• How can we ensure responsible lending and address regulatory compliance with AI.

To further exemplify these points, let’s consider what majority of the financial institutions do in the current business environment. The key question is to what extent credit decisioning processes are currently automated, how successfully we’ve adopted the latest data analytics and AI technologies, and where we can identify immediate opportunities for digital transformation. From my experience, even some of the largest global banks are not fully leveraging the available customer data. For example, while we could use customer spending data to deliver a more realistic view of their affordability and serviceability (check out our Google Cloud’s Marketplace offering as an example), most banks are still manually reviewing payslips and other physical documents for income verification. Some government-backed solutions aimed at simplifying the customers’ consent process for retrieving their bank transaction data have also proven to be inefficient or have had limited adoption.

Focusing on what the recent AI advancements mean for credit decisioning, we need to evaluate how these AI tools help us to better utilise the customers’ data and deliver real value to businesses. While LLMs enabled us to put a big step forward in providing cognitive capability to decision support systems, we are still in the early stages of building end-to-end AI Agents with the reliable reasoning capabilities to interact with the LLMs, extract historical patterns and relevant insights, and make adequate recommendations. What’s more, building and using such systems are still too expensive, time-consuming, and there are numerous regulatory and compliance implications that have not been sufficiently addressed.

If you are interested to learn more about use of AI in credit decisioning, which is a fair assumption if you have read this far, you may want to know that MOGOPLUS will be hosting an interactive event and panel discussion on this topic in Sydney on February 13th, 2025, entitled AI in Credit Decisioning … Really?

I am honoured to be joined by a panel of notable experts in the field, including Stuart Houston from Google Cloud, Des Viranna from Deloitte, and Mohammad Aman from NAB. This event will be moderated by Jennifer Harris, CPM from Sandstone Technology. I am eagerly looking forward to sharing my thoughts on how AI is going to further transform credit decisioning.

To secure your ticket, follow this link here.