KYC Automation That Clears the Clean Files First

Annual cost reduction on the retail book: $7.2M to $8.6M.

Clean applications now clear in under five minutes. Analysts see exceptions instead of volume.

A mid-size retail bank opens roughly 300,000 new customer accounts a year, and every one entered the same queue.

• One queue for every applicant. A clean file with a valid passport and a utility bill waited behind a complex one that genuinely needed an analyst.
• $40 per retail case, and days of cycle time. Across 300,000 accounts a year, on work that is mostly keying data from identity documents, proofs of address, and income records.
• Screening, verification, and risk-rating run by hand. Sanctions, politically-exposed-person, and adverse-media checks, then a risk rating, then a filed record.
• Applications get abandoned while they wait. The cost of a slow queue is not only the labour in it.
• Previous proposals died in the security review. Onboarding means processing untrusted files arriving from the open internet, into a system holding the bank's most sensitive customer data.

• Clean files clear straight through, with the rationale recorded. Anything the system is not confident about routes to a named human, so analysts see exceptions rather than volume.
• Documents are read and verified rather than keyed. Identity documents, proofs of address, and income records extracted and authenticated, with the untrusted-file handling in the tightest boundary in the system.
• Screening noise is absorbed before it reaches a person. Sanctions, PEP, and adverse-media alerts are worked the way an analyst would work them, so what arrives at the queue is worth a human.
• Beneficial owners are resolved and screened in parallel. One configured environment forks per person, so a corporate structure unfolds all at once rather than one analyst at a time.
• Every decision carries a replayable rationale. Including the clears, logged as the work happens rather than reconstructed on request.
• Customer data stays inside the bank's boundary. Control plane and storage in the bank's own region, each case in its own guest kernel, with egress allowlisted in the kernel to approved list and registry providers and nothing else.
• The regulated decision stays with a human. Every genuine hit and every low-confidence case goes to a named reviewer by policy. CreateOS is SOC 2 Type II and ISO 27001 certified.

Clean applications now clear in under five minutes, and analysts see exceptions instead of volume.

• $7.2M to $8.6M a year on the retail book. On a $12.0M line, a 40% reduction is the conservative floor we commit to in a pilot and 48% is the top of our own modelled range, not a published benchmark. Cost per case falls by around 70%.
• Under five minutes for a clean application. Down from days, with analysts seeing exceptions instead of volume.
• Recovered revenue, presented as a model not a benchmark. If compressing onboarding from days to minutes recovers just 5% of currently-abandoned retail applications, the value depends entirely on the bank's own abandonment rate and customer value, which is why both inputs get replaced with its actuals.
• A 35% to 55% straight-through rate is the honest ceiling. The agents remove the manual half of the work so the KYC team concentrates on the cases that need a human. It does not eliminate the team, and anyone who has run a KYC function knows it.

• Weeks 1 to 2, baseline. Measure the client's actual cost per case, cycle time, straight-through rate, and false-positive rate. This becomes the contract's yardstick and protects both sides in procurement.
• Weeks 2 to 6, build and integrate. Stand up the agent crew, integrate to the bank's document store, identity sources, and screening providers along allowlisted paths, deploy self-hosted inside the bank's boundary.
• Weeks 6 to 8, shadow run. Agents process real cases alongside the human team without making binding decisions. Compare, tune.
• Week 8 onward, controlled go-live. Straight-through processing switched on for low-risk retail first, humans on every exception, risk band expanding as the audit record builds.

Success criteria, agreed up front: cost per case down at least 40%, straight-through rate of 35% or better, extraction accuracy of 95% or better, 100% audit coverage of every automated decision.