For Harmoney, a direct-to-consumer digital lender in Australia and New Zealand, speed is the product. When customers apply for a loan, every extra hour increases the risk they will walk away. Four years ago, Harmoney’s average lending decision took four days. Today, it takes 17 minutes.
Harmoney turned to Camunda to automate and orchestrate the entire loan origination journey. Starting with process automation for standard applications, Harmoney expanded into agentic orchestration to tackle the thousands of creditworthy customers who were being turned away simply because the system couldn’t assess their income. The outcome is a faster, more inclusive lending experience, and a business that now converts customers it previously couldn’t serve.
Harmoney’s lending platform generated strong top-of-funnel demand, about 12,000 new applicants per month flowing through their engine. But the conversion rate told a different story: only 1,000 of those applicants were successfully funded. Eleven thousand customers left disappointed, many of them not because they were high credit risks, but because the platform could not accommodate their financial situations.
The core problem was a lack of straight-through processing. Every loan application required manual underwriting: credit officers reviewing bank statements, scrutinizing credit files, and assessing risk by hand. The average time to a lending decision was four days. In a market where speed to wallet is a genuine competitive advantage, four days was simply too slow.
A second, more stubborn challenge emerged over time. Even after automating the straightforward cases, a meaningful cohort of applicants, including those with multiple jobs or non-standard hours, could not be automatically assessed. Harmoney’s lending policy required three months of bank statements demonstrating consistent income. For applicants whose last 90 days of statements did not reflect their typical earnings, the automated system hit a wall and routed them to manual review. Before BankBot, getting a decision on these complex cases within the same day was considered lucky. Many applicants simply dropped out of the funnel entirely.
Behind the scenes, the operational reality was equally concerning. Bank statements and income documents were stored in credit officers’ email inboxes—fragmented, ungoverned, and in the words of Bobby Lei, Harmoney’s Head of Engineering, “from a security perspective, just horrific.” There was no central audit trail, no reliable way to reconstruct a lending decision if challenged in court.
“Speed to wallet is a big competitive differentiator in our industry. Four days was too slow.”
— Bobby Lei, Head of Engineering, Harmoney
Harmoney committed to Camunda Enterprise from day one—no pilot, no proof of concept—a reflection of how central they believed automation would be to their competitive position. The implementation brought the entire origination process under a single, automated flow: identity verification, income assessment, credit decisioning, and everything in between.
The impact was immediate. Decision times dropped from four days to 17 minutes and 40 seconds, roughly 90 times faster. That 17 minutes isn’t just a faster decision; it’s the time from a customer requesting $30,000 to seeing the funds in their account. Credit officers, freed from routine case reviews, could focus on the complex applications that genuinely required human judgment.
“That feeling of watching the first application go through, 12 minutes from ‘I want $30K’ to seeing it in your bank account. Particularly when customers really need the cash, having instant finance makes a huge difference.”
— Bobby Lei, Head of Engineering, Harmoney
Fast decisions solved one problem. But Harmoney was still losing thousands of customers it could have served, applicants who were creditworthy, but whose income simply couldn’t be verified through standard checks. Seasonal workers, contractors, and gig economy earners were falling out of the funnel with no alternative path forward.
BankBot changed that. Built on Camunda and powered by Google Gemini, BankBot steps in the moment a standard income check fails. Rather than routing the customer to a waiting queue, it keeps them in the flow, prompting them in real time to upload supporting documents, automatically processing those documents, and delivering a recommended decision to a credit officer within minutes. What previously took days of back-and-forth over email now takes under 15 minutes for 60–80% of cases. Customer drop-off in this cohort has fallen by 20–30%.
Importantly, every step BankBot takes is fully auditable. In a regulated industry where lending decisions must be defensible, Camunda captures a complete record of each action, giving Harmoney’s risk and compliance teams the visibility they need, without any additional effort.
BankBot launched roughly a year ago, built by a single developer over three weeks at an implementation cost of approximately $10,000. Since then, it has operated quietly and continuously, funding millions of dollars in loans that would otherwise have been lost.
The measurable outcomes are significant. Among applicants previously routed to manual review, 60–80% now receive a lending decision within 15 minutes, rather than waiting days for a credit officer to contact them. Customer drop-off across the affected cohort has fallen by 20–30%. And for the first time, sensitive financial documents are processed and stored securely in the cloud, not scattered across individual laptops and email inboxes.
The business impact of BankBot prompted Harmoney to formalize what had previously been an ad hoc retention effort. A new department, internally called Triple R (Retention, Recovery, and Reactivation) was established, with a dedicated head of retention whose mandate is to develop strategies for recapturing customers who exit the funnel. BankBot was the proof-of-concept that made the business case for this investment.
“The ROI is huge for this—kind of mind-boggling. We built it in three weeks with one developer. It’s been sitting there quietly, just funding millions of loans for us.”
— Bobby Lei, Head of Engineering, Harmoney
The security improvements are equally notable. Prior to BankBot, income documents resided in credit officers’ email inboxes with no central governance, an approach that would have been difficult to defend under regulatory audit. Today, every document is processed and stored in the cloud, with Camunda capturing a complete record of each lending decision. Harmoney’s Chief Risk Officer was, in Bobby’s words, a vocal advocate for the solution.
BankBot solved one problem in one part of the lending journey. Its success has given Harmoney the confidence to do something more ambitious: making every applicant feel like they have their own personal banker.
Today, every applicant follows the same fixed path through Harmoney’s application process, regardless of their individual circumstances. That one-size-fits-all approach means that when a customer hits a friction point, such as being asked to share their banking login credentials to verify their income, there is no alternative. They either comply or drop off.
Harmoney’s vision for the next phase is a lending journey that adapts to each customer in real time. Instead of a rigid, step-by-step process, every applicant would experience a flow tailored to their preferences and comfort level. A customer who is uncomfortable sharing their banking credentials might be offered a different way to verify their income. A customer who wants to move quickly through identity checks can. The experience feels less like filling out a form and more like speaking with someone who already understands your situation.
Projects are already underway at Harmoney to bring this to life, with initial proof-of-concepts running in different parts of the funnel. When the results are in, Harmoney expects to expand the approach across the board.
“Think of it as a virtual bank where you walk in and get your own personal banker, one that tailors the lending process to your particular situation. The interaction is very fluid. But underneath the hood is all Camunda, all process orchestration, so it gives us total auditing and the end-to-end audit trail that a highly regulated industry requires.”
— Bobby Lei, Head of Engineering, Harmoney
