What was done
Toronto fintech KOHO built Kortex AI, an internal AI platform on AWS, and pointed it at core financial operations: validating and submitting suspicious-transaction reports for anti-money-laundering work, resolving security events, letting analysts query data in plain language, checking rental leases for its Rent Reporting product, and mining customer feedback for feature requests.
Outcome
AWS's case study says KOHO now validates and submits suspicious-transaction reports three times faster, cutting days off money-laundering investigations, and security event resolution time has dropped 50%. Developer pull-request throughput rose by up to 140%, and after Kortex flagged dark mode as a top request, it was built in two days. Figures are vendor-reported.
Evidence: Vendor case study (AWS). Figures are as reported by the source and have not been independently audited by AgentsNorth.
Prompt Reconstructed by AgentsNorth
The source didn't publish its prompt. This is our reconstruction of a starting prompt for a similar workflow, not the company's actual prompt.
You are an AML analyst assistant at a Canadian fintech. Review this flagged account activity and draft a suspicious transaction report for FINTRAC. Summarize the customer profile, list each transaction with date, amount, counterparty and channel, explain which indicators make it suspicious, and note anything that argues against suspicion. Cite the source record for every fact. Mark gaps as 'needs analyst review'. Do not file anything; a compliance officer approves every report.