How Harmoney Cut Loan Decision Time 90x with Camunda

Harmoney needed to turn a slow, manually reviewed loan origination process into an instant decision engine, without losing the audit trail a regulated lender depends on.

For Harmoney, speed is the product. In direct-to-consumer digital lending across Australia and New Zealand, every extra hour a customer waits for a decision is an hour they might spend applying somewhere else. Roughly 12,000 applicants a month came through Harmoney’s platform, but only a fraction were funded, not because they were poor credit risks, but because the process built to serve them couldn’t move fast enough or flex enough to say yes.

Every application ran through manual underwriting: credit officers combing through bank statements and credit files by hand, a process that took an average of four days. Complex cases, applicants with multiple jobs, gig income, or non-standard hours, were even slower, often taking days longer while their documents sat scattered across individual inboxes with no audit trail. That gap was costing Harmoney credit-worthy customers and creating a compliance risk it couldn’t ignore.

• Committed from day one. Harmoney adopted Camunda Enterprise without a pilot or proof of concept, a signal of how central the team believed automation would be to its competitive position from the start.
• A platform built for regulated, high-stakes processes. Harmoney needed deterministic control over a lending flow that had to be defensible under audit, not just fast, which made Camunda’s mix of orchestration and governance the fit for the job.
• One flow, start to finish. Camunda orchestrates identity verification, income assessment, and credit decisioning as a single automated process, then extends into agentic orchestration through BankBot, an AI agent built on Camunda and powered by Google Gemini that handles complex income verification end to end.
• One audit trail for both phases. Whether a decision runs through the standard automated flow or through BankBot’s agentic path, Camunda captures a complete record of every step, giving Harmoney’s risk and compliance teams full visibility without extra work.

BankBot has been running for about a year, built by one developer in three weeks for roughly $10,000. In that time, it has quietly funded millions of dollars in loans Harmoney would previously have lost.

• 60 to 80% of complex cases now decision in under 15 minutes. Applicants with non-standard income, once stuck for days in manual review, get an answer almost immediately.
• Drop-off down 20 to 30% among affected applicants. Customers who previously abandoned the process while waiting on manual review now stay in the funnel.
• A defensible audit trail, by default. Sensitive financial documents are processed and stored securely in the cloud instead of sitting in email inboxes, giving risk and compliance teams the record they need.

BankBot solved one problem in one part of the journey. Its success is now pushing Harmoney toward something bigger: a lending experience that adapts to each applicant instead of forcing everyone through the same fixed path. Today, every customer follows an identical process, so a friction point, like being asked to share banking credentials, has no alternative route. Harmoney’s next phase reshapes that flow so applicants who want speed can move fast, while those with different comfort levels get another way to verify their income. Early proof-of-concepts are already running across parts of the funnel, orchestrated end to end on Camunda, with plans to expand the approach once results are in.