ThetaRay | Ray
Estimate your compliance team's capacity gains with Ray
Model your own case volumes, analyst costs, and workflows to see the hours and dollars Ray's agentic AI investigation layer can free up for your team.
1 Case volumes and performance assumptions
Enter your monthly volume and average time spent. Add or remove workflows as needed.
Select workflows
2 Analyst cost assumptions
Enter your current staffing and cost per analyst. This converts hours saved into dollars.
3 Ray impact assumption
This lever is specific to Ray — adjust it to model a more conservative or aggressive scenario.
4 Projected impact
Updates as you enter values.
Why these numbers hold up
Ray in production today
Tier 1 Bank
Phase 1 Live
Controlled go-live cleared and underway
Compliance Teams
70%
Less manual investigation effort, up to
Pilot Programs
2 Weeks
From your data to a decision
5,000 Cases / Mo
$1.6M+
Est. annual capacity savings
How this actually works
Why 70% is a real number, not a marketing line
Most of an analyst's day isn't spent making judgment calls — it's spent gathering the evidence needed to make one. Industry research on AML investigations suggests analysts spend as little as 15% of their time on actual investigative judgment; the rest is manual searching, copying, and write-up. That's the gap Ray is built to close, step by step:
Document intelligence
Ray OCRs and parses PDFs, scanned images, and spreadsheets into structured, searchable evidence the moment a case opens — the step that otherwise means an analyst opening five systems and piecing facts together by hand.
Autonomous evidence gathering
Ray pulls KYC/customer profile data, transaction history, counterparty relationships, and adverse media automatically — the exact searches an analyst would otherwise run one at a time, across disconnected tools.
Narrative drafting & correlation
Ray synthesizes what it gathered into a structured investigation narrative, flags contradictions or anomalies, and cross-references prior alerts — turning a blank page into a draft the analyst edits instead of writes from scratch.
Analyst review, not autopilot
Every draft is reviewed, edited, and finalized by a human before it's actioned. Ray removes the manual legwork; it doesn't remove the judgment call — which is also why the output holds up under audit.
See it on one case
Take a sanctions name-screening hit on an existing correspondent-banking customer. In a typical 20-minute review today, an analyst spends roughly 15 of those minutes pulling the customer's KYC file, checking the screening hit details, scanning transaction history for context, searching adverse media by hand, and checking whether the same customer has prior alerts on file — all before writing a word of the disposition. Ray runs that first pass automatically: KYC pulled, transactions summarized, adverse media scanned, prior alerts cross-referenced, and a draft disposition written with every fact sourced back to where it came from. The analyst opens a case that's already assembled and spends their time confirming the call, not building the evidence for it.
This is also why the number above moves with your own inputs rather than holding at a fixed multiple: the time Ray gives back scales with how much of a case today is manual evidence-gathering and write-up versus judgment — heavier, document-heavy workflows (EDD, SAR drafting) see more of it than a quick sanctions disposition.
Estimates are directional, based on ThetaRay published benchmarks and the assumptions entered above — not a contractual commitment.