Reduced loan decisions from 24 hours to 15 minutes with Gemini
Decreased applicant drop-off rates by 20% to 30%
Deployed production-ready AI in three weeks on Google Cloud
Secured up to 20,000 monthly financial documents on Google Cloud
Harmoney uses Gemini on Google Cloud to slash loan funding to 15 minutes, cutting drop-offs and accelerating personal finance.

We realised we still couldn't out-code the 'blunt tool' problem. Code follows rigid rules, but financial life is complex and personal. We needed an agentic solution that could reason through financial data at the speed of a 100% online application.
Marija Stulich
Head of Retention, Harmoney
As a direct-to-consumer digital lender operating across Australia and New Zealand, Harmoney relies on speed and personalisation to stand out. Specialising in unsecured personal loans and automotive financing, the company’s business model depends on frictionless, hyper-personalised risk-based pricing.
But fulfilling their promise of quick funding was hindered by the complex nature of modern financial lives. To comply with responsible lending standards, Harmoney has to accurately assess customer bank statements. Historically, the company relied on third-party categorisation software.
These deterministic, rule-based systems functioned effectively as ‘blunt tools’ for standard 9-to-5 salaries but failed to properly interpret the nuance of side hustles or irregular income streams.
It triggered a prolonged back-and-forth communication loop involving at least two customer service representatives and a credit officer. Consequently, getting a loan decision often took over 24 hours: a delay that led to significant customer drop-off.
Harmoney recognised the need for a probabilistic reasoning engine and selected Google Cloud to build an agentic architecture capable of evaluating complex financial data with the speed required for a 100% online application.
Bank Bot isn’t a chatbot. It’s a backend AI powerhouse built on Google Cloud that extracts information, classifies documents, and reasons just like a human credit officer.
Marija Stulich
Head of Retention, Harmoney
Leveraging the existing security and compliance guardrails on Google Cloud, Harmoney introduced ‘Bank Bot,’ a sophisticated back-end generative AI agent built on Gemini platform.
Unlike standard customer-facing chatbots, Bank Bot operates silently within Harmoney’s Stellare 2.0 platform. It instantly classifies, extracts, and reasons over unstructured bank statements, automating affordability checks without human intervention.
As Harmoney built this solution on a unified cloud foundation, it avoided multi-vendor complexity. Because their security and compliance pipelines were already established in Google Cloud, a hyper-lean team consisting of just one developer and one product manager was able to spend 90% of their time on strategy and ideation, and only 10% on the actual build. The business impact was transformative. By introducing generative reasoning into the pipeline, Bank Bot reduced the average loan decision time from 24 hours down to just 15 minutes.
Today, more than one in four applications are fully funded within an hour, and the fastest end-to-end loans are fully funded in less than 10 minutes. Crucially, this seamless experience slashed applicant drop-off rates by 20% to 30%.
Bank Bot also securely processes up to 1,000 highly sensitive documents a month within a fully sovereign, audited pipeline, ensuring customer data never sits in unprotected silos like email inboxes. In removing the manual data extraction loop, Harmoney has empowered its credit officers to focus on empathy-led decision-making, driving stronger relationships and significantly increasing repeat business.

Harmoney is a fast-growing, direct-to-consumer digital financial services provider offering hyper-personalised personal loans, automotive finance across Australia and New Zealand.
Industry: Financial Services, Software and Internet
Location: New Zealand
Products: Google Cloud, Gemini, Gemini platform