Harmoney
90x Faster Lending Decisions: How Harmoney Scaled Loan Approvals With Agentic Orchestration
Harmoney cut loan approval time from 4 days to 17 minutes using Camunda process orchestration, then deployed BankBot to serve complex income cases — reducing customer drop-off by 20–30%.
90x Faster Lending Decisions: How Harmoney Scaled Loan Approvals With Agentic Orchestration
Harmoney is a direct-to-consumer digital lender in Australia and New Zealand where speed matters significantly. Customer loan applications previously took four days for approval; this has been reduced to 17 minutes. Camunda automated and orchestrated their complete loan origination workflow.
Harmoney received approximately 12,000 new applicants monthly, but only converted 1,000 into funded loans. The remaining 11,000 were rejected not due to credit risk, but system limitations. Manual underwriting caused delays; every application required credit officers reviewing statements and assessing risk manually. Four-day decisions proved too slow competitively.
A secondary challenge emerged: applicants with multiple jobs or irregular income couldn't be automatically assessed. Policy required three months of consistent bank statements. Those whose recent statements didn't reflect typical earnings hit system walls and required manual review, causing many to abandon applications.
The agent is the brains behind it, but the hands — that's when Camunda comes in. Camunda's able to reach out to the customer, take that document, and orchestrate the entire workflow. That deterministic behavior is really important in lending. We used Camunda as our audit trail.
Bobby Lei
Head of Engineering, Harmoney
Harmoney committed to Camunda Enterprise immediately without pilots or proof-of-concepts. The implementation unified the entire origination process: identity verification, income assessment, credit decisioning. Decision times dropped from four days to 17 minutes 40 seconds — approximately 90x faster.
BankBot, built on Camunda and powered by Google Gemini, intervened when standard income checks failed. Rather than queuing applicants, it maintained engagement by prompting document uploads, automatically processing submissions, and delivering recommendations to credit officers within minutes. Previous multi-day email exchanges now resolved in under 15 minutes for 60–80% of cases.
Every BankBot action remained fully auditable, critical in regulated lending where decisions must be defensible. Camunda captured complete action records, providing risk and compliance teams necessary visibility without additional effort.
