Product updates, partnerships, case studies, and ideas on AI-powered compliance operations — from the team building it.
There's a new analyst on the team. Justin runs KYB and KYC investigations end to end — ownership, related parties, deep web research, and screening — ending in a decision-ready case file.
Get The Alert Queue every other week. Unsubscribe anytime.
FairPlay tested Axle's Baley screening agent on 1,048 hard sanctions, PEP and adverse media cases: 97.5% accuracy, 99.2% precision. Nobody required it. We asked for it.
AlterBank is taking Rose from pilot to full production after cutting investigative labor costs by more than 70% — while its existing compliance team processed three times more AML alerts without adding headcount.
Justin joins the team as Axle's KYB & KYC diligence agent, Copilot lands in every investigation, and Rose gains semantic search, rule-based dispositions, and on-demand investigations.
Axle Copilot sits alongside every open alert, investigation, and diligence case — answering questions, rerunning investigations with new context, and taking action behind plan cards and predicted-impact previews.
Justin is Axle's diligence agent — he runs KYB and KYC investigations end to end, from ownership and related-party research to a decision-ready case file with a suggested disposition and next steps.
Axle's integration with Chainalysis brings blockchain risk signals, exposure data, and on-chain context directly into Rose's transaction monitoring investigations — no tab-switching, no manual lookups, and a full audit trail on every disposition.
The Axle MCP server lets compliance teams operate Rose, Baley, and Sam from MCP-aware tools — pull live alert metrics, trigger reruns, and inspect investigation context without leaving your own AI stack.
Rose gains a Chainalysis integration, API V2, end-to-end case management, custom PDF reporting, and transparent flag-based risk assessments; Baley adds business-name matching and SOE classification controls.
Case management and OneSchema CSV uploads land in Rose, while Baley gains configurable PEP handling, location matching, and multi-point approvals — plus deeper audit trails across the board.
Hundreds of updates hit production this quarter — OSINT and API upgrades for Baley, blockchain investigations and multi-institution SOPs for Rose, and Sam officially out of beta.
Rise put Rose behind its transaction monitoring queue — investigating alerts end to end in its existing stack, 24/7, with a documented rationale on every disposition. Here's how a review runs and what changed for the team.
We're launching the Axle blog — and kicking it off with a look at what we shipped during our internal hackathon: global search across Axle, a redesigned transactional email system, and this blog itself.
How to run a good guy list in sanctions screening that survives an audit: what to suppress, expiry, re-review on list changes, and logging.
How transliteration breaks sanctions screening for Arabic, Cyrillic and Chinese names, and the script-specific rules and test cases that fix it.
When a fund opens an account, who do you verify? The CDD Rule's pooled vehicle relief, the control prong for GPs, and when to look through to LPs anyway.
Screening stops a payment before it moves; monitoring finds patterns after. Owners, SLAs, alert types, and where AML teams blur the two controls.
The SAR 30-day clock starts at initial detection, not the alert date. What starts it, when 60 days applies, continuing-activity timing, and how to document it.
The 16 entity types excluded from FinCEN's beneficial ownership rule, the control-only cases, account exemptions, and where KYB teams get it wrong.
FinCEN's 2026 program proposal backs moving AML resources away from low-risk activity. The decision record that makes a reallocation survive an exam.
What an MSB AML program must contain under 31 CFR 1022.210: Form 107 timing, FIN-2016-G001 agent oversight, the $2,000 SAR threshold, independent review.
Stablecoin transaction monitoring: USDT on Tron, guarantee marketplaces, on- and off-ramp rules for banks and exchanges, and the GENIUS Act date.
Above the line below the line testing for AML thresholds: worked example with sampling math, productive rates by band, and a tuning memo examiners accept.
How to build an AML investigation quality assurance program: risk-tiered sampling, scorecard fields, calibration, drift triggers, and AI agents.
Running credit union BSA compliance with a small team: NCUA 748.2 minimums, what to automate first, exam prep, and sharing resources through CUSOs.
Twelve BSA/AML board report KRIs with definitions, triggers and trend logic, a one-page quarterly template, and what to leave out of the board pack.
A 10-day playbook for turning a FinCEN alert into monitoring rules, SAR key terms, a lookback, training and a board memo, using the 2026 scam-center alert.
There's no regulatory benchmark for alerts per analyst per day. Here's how to calculate your own from alert mix, tiering, QA and handle time, and defend it.
A BSA independent testing checklist built from the FFIEC manual: scope, the files auditors request, how alert and SAR decisions get tested, and evidence staging.
What the GENIUS Act and FinCEN's 2026 PPSI proposals require of stablecoin issuers, and the AML, CIP and freeze controls to build before Jan 2027.
How to monitor FedNow, RTP and Zelle for AML and fraud: limits, request-for-payment abuse, pre-send vs post-send controls, and real alert SLAs.
The real difference between AML and fraud teams, what a FRAML model should share (data, cases, 314(b)) and what to keep separate (SAR calls, metrics).
What to look for in an AI vendor's SOC 2: Type 2 scope, carved-out LLM providers, CUECs, retention and training terms, and the model risks a SOC 2 skips.
What sponsor bank fintech AML oversight needs in 2026: KRIs, ledger and alert access, SAR ownership, and what the OCC's CFSB consent order required.
How to run a 314(b) information sharing process after FinCEN's June 2026 fact sheet: fraud, real-time sharing, request templates, logs, and joint SARs.
Money mule red flags from FinCEN advisories, the baseline-shift and in-fast/out-fast patterns that catch them, and a worked SAR narrative example.
Perpetual KYC vs periodic review: what the FFIEC actually requires, a trigger taxonomy, false-trigger tuning, and how to run both during transition.
OWASP LLM01 translated for BSA officers: which inputs your AML agent reads, what tools it can call, where humans gate it, and the test evidence to ask for.
What a fintech's first compliance hire should own, outsource, and automate, how senior to hire, and what the sponsor bank asks for in month one.
Transaction monitoring scenario documentation examiners can follow: a one-row-per-rule register with typology, logic, thresholds, owner, tuning, and results.
SAR vs STR difference for US teams expanding into Canada: thresholds, deadlines, narrative standards, portals, and Bill C-12 penalty changes.
An AI-generated pre-screen is defensible to a regulator when you can show its logic, reproduce each decision, sample its closures, and keep the evidence.
The 314(a) request process step by step: what to search, clearing name matches, logging negatives, and what a positive match does and doesn't require.
Who decides, how to document it, SAR timing, notice language, returning funds, and what the 2026 reputation-risk rule means for exit memos.
Block vs reject, the 10-business-day OFAC reports, the September 30 annual report of blocked property, and the 10-year recordkeeping rule.
How to design an expected account activity questionnaire that gives transaction monitoring a usable baseline, and the onboarding fields to drop.
CTR aggregation rules, Phase I/II exemptions and DOEP filings, late-filing fixes, and why exempt customers still need SAR monitoring.
A four-test materiality filter for adverse media hits, a category-by-stage escalation table, and what a defensible dismissal note needs.
The CIP rule asks for three data points on a business. Here is what FinCEN requires, what CDD adds, and where banks and fintechs over- and under-collect.
What an internal suspicious activity referral form should capture, what frontline staff should never write, and how referrals become alerts and SAR cases.
How payments watchlist screening differs from customer screening: which fields to screen, latency budgets, false-positive tolerance, and instant rails.
KYB for trusts under FinCEN's CDD Rule: when the trustee is the beneficial owner, what to do with corporate trustees, and where risk-based diligence goes further.
How long must a SAR be retained? Five years from filing. The full BSA retention map for SARs, CTRs, CIP and BO records, plus OFAC's 10-year rule.
When a date of birth or nationality mismatch can clear a sanctions screening hit, when it can't, and how to document the decision defensibly.
How to build an ownership chart for a complex, multi-layer entity: edge-list data, CDD 25% math through layers, OFAC 50% math, and when to stop.
PEP status alone is not a SAR trigger. The red flags that are, the FinCEN key terms to use, and how to write the narrative for a PEP SAR.
How the OFAC 50 percent rule works on aggregate and indirect ownership, and what OFAC's March 2026 sham-transaction advisory adds to KYB and screening.
How to build a customer risk rating methodology examiners accept: factor selection, weighting, floors that stop averaging-out, and back-testing.
A SAR quality review checklist for second reviewers: what to check before filing, what not to re-litigate, and how to feed QA findings back upstream.
A practical AML risk assessment template: inherent risk, controls, residual risk, and action plan, with a worked example and what FFIEC and FinCEN expect.
General chatbots can summarize a case file. They can't pull evidence, keep an audit trail, or stay inside your controls. Here's where the line sits.
A field-by-field AML escalation memo template for L1-to-L2 and L2-to-SAR handoffs, with examiner expectations and the six-month cold-read test.
US rules don't define PEPs or require PEP screening. Here's what FinCEN and the FFIEC expect for PEP CDD, and how that differs from EU and UK law.
A six-section EDD memo template (trigger, profile, SoF/SoW, findings, residual risk, approval) with an annotated example and reviewer pushback.
What the Sept 2026 joint statement lets banks tell customers about fraud and account closures, the subpoena rule, and front-line scripts.
A weighted scorecard for comparing watchlist screening solutions: list coverage, matching logic, secondary identifiers, alert volume, latency, audit trail.
FIN-2026-R001 ends beneficial ownership re-collection at every new account. What to change in your CDD procedures, what to keep, and how to document it.
FinCEN's FIN-2019-A003 lists 23 CVC red flags. Which ones you can code as monitoring scenarios for exchanges and banks, plus the 2025 kiosk update.
A day 1 to 90 checklist for a new BSA officer: what to pull, who to meet, what to fix first, and what to put in your first board report.
Onboarding screening proves a customer was clear on day one. Watchlist monitoring proves they still are: three triggers, scope, and what examiners expect.
FinCEN's October 2025 FAQs say you need not document a decision not to file a SAR. Why you still should, and how much is enough at each tier.
A TPRM-grade AI vendor due diligence questionnaire for AML teams: model change control, data and training use, subprocessors, and evidence access.
A working set of alert disposition codes for AML teams: false-positive sub-reasons, true positives, escalations, data errors, and how codes feed tuning and QA.
The FinCEN SAR key terms list as of September 2026: every 2026 term, where it goes on the SAR, which pair with a field 34 code, how to keep it current.
FinCEN ended BOI reporting for U.S. companies in August 2026. What a beneficial ownership certification still requires, and when to refresh it.
SR 26-2 excludes generative and agentic AI from model risk guidance but still expects governance. Compliance AI needs: inventory, validation, monitoring.
Which watchlists to load, how fast list updates must reach your filter, and how delta screening rescreens only what changed without losing the audit trail.
A PEP check flags customers who hold or held prominent public roles, plus family and close associates. Who counts, which tiers to screen, what banks do.
Five annotated SAR narrative examples, from structuring to crypto off-ramps, mapped to FinCEN's five Ws with the key terms and fields each typology needs.
FinCEN's FIN-2026-Alert005 lists 16 scam center and pig butchering red flags. Which ones banks can detect, how to code them into rules, and filing the SAR.
The control prong means one senior person who controls a legal entity, such as a CEO, CFO or managing member. How it differs from the 25% ownership prong.
A staged AML program modernization path from rules-based monitoring to AI agents, built for FinCEN's 2026 effectiveness standard and the revised SR 26-2.
What agentic AI means for compliance teams: how agents differ from rules and models, where regulators landed in 2026, and the controls that survive exams.
AML fines 2026: what FinCEN, DOJ, OCC and NYDFS actions against UBS, Canaccord, EagleBank, OKX and Block reveal about the failures regulators punish.
Outsourced AML alert review, in-house analysts, or AI agents? A practitioner comparison of cost, quality drift, accountability, and hybrid models.
FinCEN's eight AML/CFT priorities are unchanged in 2026, but the April proposed program rule changes how you must use them. What BSA officers should do now.
A practitioner's map of the AML compliance tech stack in 2026: the six core categories, where AI agents fit, and what April's rule changes mean for each layer.
A step-by-step walkthrough of AI transaction alert investigation: data gathering, typology analysis, disposition rationale, human review, and what regulators expect.
SR 26-2 leaves generative and agentic AI out of scope. What model validators still need before approving AI in AML, and a practical validation package.
How to cut KYB onboarding friction without adding AML risk: FinCEN's 2026 CDD relief, FATF's risk-based approach, and a tiered verification workflow.
Which transaction monitoring metrics matter: how to measure alert precision, recall, and SAR yield — and what examiners expect under FFIEC and SR 11-7.
What data an AI transaction monitoring pilot needs — transactions, KYC, alert dispositions — plus what SR 11-7 and the 2018 joint statement expect.
How automated SAR narrative generation works, what FinCEN expects from narratives, and a workflow that keeps analysts accountable for the why.
How to turn screening alert adjudication history into a feedback loop that cuts repeat false positives without weakening sanctions coverage.
Which transaction monitoring scenarios still earn their alert volume — and a defensible process for tuning or retiring the ones that don't.
How often to refresh KYC: no rule sets a fixed interval. Common cycles by risk tier, trigger events that force early review, and how to defend it at exam.
Test an AI agent against your analysts before go-live: golden datasets, blind parallel runs, override analysis, and the evidence model risk teams ask for.
Where real-time sanctions screening is non-negotiable, where batch rescreening fits, and how to run both without drowning in post-list-update alert spikes.
How to calibrate transaction monitoring thresholds by risk segment: what FFIEC examiners expect, a five-step tuning workflow, and the documentation that survives an exam.
FinCEN data puts annual SAR volume near 4.7M. How to benchmark your SAR volume with ratio metrics — and what examiners actually look for.
How mid-size bank compliance teams evaluate Actimize alternatives: replace, wrap, or augment — plus the regulatory guardrails for AI-based monitoring.
Verify a business registration in any state: run the Secretary of State search, confirm good standing, check the registered agent and EIN, spot red flags.
A BSA officer's framework for build vs buy compliance AI: what regulators expect, the true cost of building, and when each path makes sense.
How to structure a transaction monitoring investigation narrative: the five W's, flow-of-funds detail, disposition rationale, and SAR-ready writing habits.
Cut sanctions screening alert noise without loosening match thresholds: data hygiene, suppression rules, weak aliases, and AI-assisted alert review.
Crypto transaction monitoring spans two ledgers — fiat rails and on-chain activity. How alerting, red flags, and investigations change, per FinCEN and FATF.
How to build an AI decision audit trail that holds up in an exam: what to log per decision, SR 11-7 documentation, and making agent reasoning reviewable.
Shell company red flags from FATF and FinCEN: formation, ownership and transaction signs to check at onboarding and refresh, and how to act on them.
How to build an audit-ready SAR filing workflow: deadline clocks, narrative standards, supporting documentation, and where automation fits.
How watchlist screening works: which lists to screen against, how to evaluate vendors, and how to build an alert workflow that survives an exam.
What causes AML alert backlogs, what regulators expect when you fall behind, and a risk-ranked plan to clear the queue without hurting SAR quality.
How human-in-the-loop AML works in practice: which decisions automate safely, where analyst judgment stays mandatory, and what regulators expect.
What separates source of funds from source of wealth, when EDD requires each, and how to corroborate both without stalling onboarding.