top of page
Website Quantum Payments (1).png

The 60-Minute Alert: Why Compliance Speed Is Becoming a Payment Performance Metric

Writer: Quantum Payments
Quantum Payments
Sep 18
8 min read

When dLocal said on 17 September that a Level 1 compliance alert can take up to 60 minutes to investigate, it highlighted more than an AML workload problem.

It exposed a payments performance problem.

dLocal’s partnership with Oscilar is designed to extend AML monitoring, sanctions screening and case management across more than 60 markets and 1,000 alternative payment methods. The objective is to use agentic AI to assemble evidence and manage routine investigation work, while human analysts retain final decision-making authority.

The important shift is conceptual: compliance queues are payment queues.

Every false positive can delay merchant onboarding, hold a payout, interrupt a cross-border transfer or slow a refund. Financial crime compliance is no longer simply a cost centre operating beside payments. Its accuracy and response time increasingly determine whether payments flow at the speed a customer, merchant or marketplace expects.

False positives are now a payment performance problem

A false positive occurs when a legitimate customer, transaction or counterparty is incorrectly flagged as suspicious. In AML transaction monitoring and sanctions screening, this is often caused by common names, insufficient context, rule thresholds or fragmented data.

The impact is operational:

  • A merchant waits longer to complete KYC or KYB onboarding.

  • A newly approved merchant starts with lower transaction limits.

  • A legitimate payout is held while an analyst reviews an alert.

  • A refund misses its expected service window.

  • A cross-border payment waits for manual investigation or additional information.

  • Analysts spend time clearing routine noise instead of examining genuine risk.

  • Payment providers become more conservative with corridors,methods or correspondent relationships.

This is closely related to the approval accuracy problem explored in our analysis of false declines, AI fraud detection and payment success rates. A payment can fail because a fraud model is too aggressive, or because a compliance process cannot distinguish a legitimate transaction from a risky one. In both cases, the result is friction that the customer experiences as a failed payment.

The cost is also difficult to see in a conventional compliance dashboard. A team may report alert volumes and closure rates while the commercial business absorbs longer onboarding times, lower payment success rates and more customer support contacts.

Neon network graph mapping fund flows with one anomalous hub glowing red

Why the decision window has collapsed

Payments are moving faster while risk signals are becoming more complex.

Real-time account-to-account rails, instant onboarding, stablecoin settlement and embedded finance compress the time available to make a safe decision. Agentic commerce adds another layer: software may initiate or manage a payment on behalf of a person or business, creating new patterns that legacy systems were not designed to recognise.

ComplyAdvantage’s State of Financial Crime 2026 found that 61% of firms prioritise real-time monitoring. Yet 67% of teams still take around 15 minutes to clear a single alert, and 81% of North American firms take more than five minutes per sanctions alert. More than half of respondents also reported managing eight to ten screening tools.

That creates a structural mismatch. The payment may settle in seconds, while the exception process operates in minutes or hours.

The answer is not to remove controls. It is to make controls more contextual. A provider that treats every automated payment as suspicious will create unnecessary friction as agentic payments scale. This is why the 10 September collaboration between Visa, Mastercard and Ant International on Know-Your-Agent interoperability matters. Agent identity and trust need to become usable compliance signals, not just additional fields that generate alerts.

Our earlier discussion of the Know Your Agent trust layer examines this emerging requirement in more detail.

What agentic AI changes inside the compliance stack

Agentic AI is not simply a faster rules engine. Properly governed, it can perform a sequence of tasks across the investigation workflow:

  1. Alert triage: prioritising cases by risk, urgency and customer context.

  2. Evidence assembly: retrieving KYC data, transaction history, customer profiles and relevant external information.

  3. Entity resolution: determining whether a name, account or sanctions match relates to the correct person or organisation.

  4. Network analysis: identifying connected accounts, shared addresses, fan-in/fan-out hubs and circular transfers.

  5. False-positive resolution: explaining why a flagged activity is consistent with expected behaviour.

  6. Escalation: sending ambiguous or high-risk cases to a human reviewer with the relevant evidence already assembled.

  7. Auditability: recording the data, sources, reasoning and policy applied to each decision.

Recent examples indicate the potential. In testing by Alviere, Hawk’s AML Investigative Agent escalated every alert that analysts had confirmed as a true positive and recommended closure for 98% of alerts previously classified as false positives. It also identified additional suspicious patterns, including a 400-times increase in transfer size linked to a fan-in/fan-out hub.

A SymphonyAI proof of concept with a global payments processor reported tenfold faster alert processing, around a 90% reduction in manual adjudication effort and a 99% reduction in false positives in a sanctions workflow.

The opportunity is significant, but the execution gap remains. ComplyAdvantage reports that 93% of firms use, pilot or evaluate AI for customer screening and 87% do so for transaction monitoring. However, only about 33% and 32% respectively use advanced agentic or predictive AI. Cost outcomes also vary: Bain found that around 40% of firms achieved only 0–10% cost impact from AI, despite 37% targeting reductions of 11–20%.

The lesson is that automation alone is not the strategy. Agentic AI must be connected to reliable data, clear controls and measurable payment outcomes.

Deloitte’s “auditable bank” analysis makes the governance requirement clear. Firms need decision-level traceability, guardrails, escalation paths, evidence surfaces, data lineage and ongoing assurance for performance drift and controls testing. Human analysts should move towards judgement and oversight rather than disappear from the process.

What this means for merchants and payment operators

Merchants should treat compliance speed as part of provider selection.

Ask how quickly a provider can complete onboarding, how long a held payout typically remains pending and whether refund delays can be separated from scheme or banking delays. Ask whether performance is measured by jurisdiction, payment method, customer segment and risk tier.

Payment operators should connect compliance data with payment orchestration, settlement and reconciliation. A risk decision that is invisible to the operations team is difficult to manage. A unified view can show whether a payment was declined, held, rerouted or delayed because of a compliance control.

This is the role of a broader unified Quantum Payments platform, where payment orchestration, operational data, accounting and business intelligence can work together. Compliance should not sit in a disconnected system that hides its effect on commercial performance.

Merchant and payment operator checklist

Use these questions to establish whether compliance is supporting or constraining payment performance:

  • Measure alert acknowledgement, investigation and resolution SLAs, including alert ageing.

  • Track onboarding-to-first-payment time by market and merchant risk tier.

  • Report payment holds, payout delays and refund delays caused by compliance review.

  • Measure false-positive rates by jurisdiction, payment method and customer segment.

  • Set separate targets for screening latency, manual review time and final case resolution.

  • Require vendors to provide explainability, decision evidence, data lineage and audit logs.

  • Define human-in-the-loop rules, including who can approve, override or escalate an AI recommendation.

  • Classify trusted agent traffic rather than treating all automation as suspicious.

  • Connect compliance reporting with payment orchestration, settlement and reconciliation.

  • Test models and workflows for drift, changing typologies and market-specific regulatory requirements.

Frequently asked questions

What is a false positive in AML transaction monitoring?

A false positive is an alert generated against legitimate activity that is incorrectly treated as suspicious. It may result from a name similarity, an incomplete customer profile, an overly broad rule or a lack of transaction context.

Why do false positives matter to payment performance?

They consume analyst capacity and create unnecessary holds. That can delay onboarding, payouts, refunds and cross-border settlement while lowering straight-through processing and customer satisfaction.

How can agentic AI reduce AML false positives?

Agentic AI can assemble evidence, resolve entities, compare behaviour with a customer’s expected profile and analyse connected accounts before recommending closure or escalation. This allows routine alerts to be resolved more consistently while directing human attention to higher-risk cases.

Does agentic AI remove human analysts?

No. In a responsible deployment,AI handles repetitive investigation work while human reviewers retain decision authority for defined cases. Analysts also oversee controls,challenge outputs and investigate complex or novel risks.

What should merchants ask their payment provider about compliance?

Ask for onboarding and payout SLAs, hold and refund metrics, false-positive rates by market, explainability standards, escalation processes and reporting integration. Merchants should also ask how the provider identifies legitimate agent-initiated payments and protects them from unnecessary friction.

Compliance precision is becoming a competitive payment capability

The 60-minute Level 1 alert is a useful warning. It shows how a compliance process can become a throughput constraint even when the underlying payment infrastructure is fast.

The providers that win will not be those that choose speed over control. They will be those that combine accurate financial crime compliance with payment orchestration,contextual decisioning and auditable automation.

In the next phase of payments,compliance precision will influence who can onboard faster,settle more reliably and expand into new markets with confidence. Compliance will not merely protect payment performance. It will help define it.

Authoritative sources

Daily handover to Sonny

Wix CMS slug: /post/compliance-speed-payment-performance-aml-false-positives

Publication target: Friday, 18 September 2026, 7:30am AEST, so both social posts go out after it is live.

LinkedIn post 1

Scheduled: 8:08am AEST

Angle: Data-led strategic hook. Compliance queues are payment queues, using the 60-minute Level 1 alert figure from dLocal’s Oscilar partnership, the 15-minute average alert clearance figure and the 65–85% false-positive resolution potential from agentic AI.

Exact copy:

A compliance alert is no longer just a compliance task.

It is a payment-performance event.

dLocal says a Level 1 alert can take up to 60 minutes to investigate.

ComplyAdvantage reports that 67% of financial crime teams still take around 15 minutes to clear a single alert.

Meanwhile,agentic AI could autonomously resolve 65–85% of false positives when deployed with the right controls and human oversight.

That gap affects:

→ Merchant onboarding → Cross-border payment clearance → Payout reliability → Refund speed → Payment success rates → Analyst capacity

The strategic question for payments leaders is changing.

Not: “How do we reduce AML cost?”

But: “How does compliance accuracy set our payment performance?”

Relevant tags: @dLocal, @Oscilar, @ComplyAdvantage, @Quantum Payments

Visual concept: Abstract neon compliance queue with multiple alert paths converging into one glowing decision core. Use the hero image. Overlay text: “Compliance queues are payment queues”.

First comment:

If 15 minutes is the average time to clear an alert, which payment metric is most exposed in your operation: onboarding, payouts, refunds or cross-border settlement?

LinkedIn post 2

Scheduled: 3:23pm AEST

Angle: Operator and merchant-operational checklist. Focus on readiness for measuring and governing compliance-driven payment delay.

Exact copy:

Is your compliance operation slowing payments without appearing in your payment dashboard?

Run this checklist:

☐ Measure alert ageing and time-to-resolution ☐ Track onboarding-to-first-payment time ☐ Report compliance-driven payout holds ☐ Measure refund delays caused by review ☐ Segment false positives by market and payment method ☐ Require explainable AI decisions and audit logs ☐ Define human approval and escalation points ☐ Identify trusted agent-initiated payment traffic ☐ Connect compliance data to orchestration and reconciliation

The practical goal is not fewer controls.

It is better controls that protect payment speed while keeping human judgement where it matters.

Relevant tags: @Quantum Payments,@Deloitte,@SEON,@ComplyAdvantage

Visual concept: Neon payments timeline showing onboarding,screening,settlement and payout gates connected by one continuous light stream. Use the second inline timeline image. Overlay text: “Measure every compliance-driven delay”.

First comment:

Operators:start by pulling one week of data and asking a simple question:how many minutes of payment delay were caused by compliance review rather than the payment rail itself?

 
 
bottom of page