Picture the compliance analyst's morning: 400 alerts in the queue. By midday, 380 of them are false positives — wrong thresholds, misconfigured rules, noise. The other 20 need a closer look.

Now picture a structuring scheme running through those same accounts. No single transaction looks wrong. No individual deposit hits the reporting threshold. The customer's behaviour matches dozens of legitimate customers. The pattern only exists if you look across 14 accounts over 11 weeks — which nobody did, because the queue had 400 alerts in it.

That is why structuring is the hardest form of financial crime to catch. It is not poorly hidden. It is built to be invisible.

What Structuring Is and How Smurfing Differs

For a full definition, see the Tookitaki glossary entry on smurfing. This article focuses on detection and reporting.

The short version: structuring means deliberately breaking up transactions to stay below regulatory reporting thresholds. One person depositing AUD 9,500 on Monday, AUD 9,800 on Wednesday, and AUD 9,300 on Friday — instead of a single AUD 28,600 deposit — is structuring. The intent is to avoid triggering a threshold reporting requirement, and that intent is the offence.

Smurfing is the same offence executed through multiple people. Rather than one person making repeated sub-threshold deposits, a network of individuals — "smurfs" — each make smaller deposits into the same account or a connected set of accounts. The underlying goal is identical: aggregate the cash while keeping each individual transaction below the reporting radar.

Both are placement-phase techniques within the three stages of money laundering. What makes them particularly difficult is that the individual transactions, viewed in isolation, are entirely legitimate.

Ten Red Flags That Signal Structuring

These red flags are not individually conclusive. They are indicators that warrant escalation to a Suspicious Matter Report or Suspicious Transaction Report when found in combination.

1. Repeated cash deposits just below the local reporting threshold

The clearest signal. A customer depositing AUD 9,400, AUD 9,700, and AUD 9,200 across three weeks is staying intentionally below Australia's AUD 10,000 cash transaction reporting threshold. The same pattern in Singapore sits below SGD 20,000; in the US, below USD 10,000.

2. Multiple transactions on the same day at different branches

A customer making three separate cash deposits at three different branch locations on the same day — each below threshold — cannot plausibly be explained by convenience. Branch diversity exists to avoid system-level aggregation.

3. Round-number deposits slightly below threshold

Real cash transactions tend to be irregular amounts. Deposits of exactly SGD 19,900, SGD 19,950, or SGD 19,800 — consistently round and consistently just under SGD 20,000 — suggest deliberate calculation rather than organic cash flow.

4. Shared identifiers across multiple accounts making similar deposits

When several accounts share a phone number, residential address, or email address, and each account is receiving sub-threshold cash deposits at similar intervals, the accounts are likely part of a structured network rather than unrelated individuals.

5. Accounts with no other activity except periodic sub-threshold cash deposits

A bank account that receives a cash deposit of AUD 9,800 every two to three weeks — and does nothing else — has no plausible retail banking purpose. Dormancy broken only by structured deposits is a strong indicator.

6. Rapid cycling: deposit, transfer, withdrawal in quick succession

Cash arrives, moves to a second account immediately, and is withdrawn within 24 to 48 hours. The rapidity defeats the logic of ordinary cash management and suggests the account is a pass-through in a structuring chain.

7. Multiple third parties depositing into the same account

Three different individuals — none of whom is the account holder — making cash deposits into the same account within a short window is the operational signature of smurfing. The account holder is coordinating a network of smurfs.

8. New accounts with immediate high-frequency sub-threshold activity

An account opened less than 30 days ago that immediately begins receiving several sub-threshold cash deposits per week has not developed an organic transaction history. The account was opened for the structuring activity.

9. Mule account patterns

The account receives multiple small deposits from various sources, accumulates the balance, then transfers the full amount to a single destination account. The collecting-and-forwarding pattern is a textbook mule structure.

10. Timing clusters at branch opening or closing

Transactions concentrated in the first 15 minutes after branch opening or the last 15 minutes before closing can indicate coordination — perpetrators managing detection risk by limiting teller exposure or taking advantage of shift-change gaps in oversight.

APAC Reporting Obligations: Thresholds and Timeframes

Compliance officers across the region operate under different regulatory frameworks. These are the current obligations as of 2026.

Australia — AUSTRAC

Under the Anti-Money Laundering and Counter-Terrorism Financing Act 2006:

• Threshold Transaction Report (TTR): Required for all cash transactions of AUD 10,000 or more, or the foreign currency equivalent. Must be submitted to AUSTRAC within 10 business days.
• Suspicious Matter Report (SMR): Where a reporting entity forms a suspicion that a transaction or customer may be connected to money laundering, financing of terrorism, or proceeds of crime, the SMR must be submitted within 3 business days of forming that suspicion (or 24 hours if terrorism financing is suspected).

Structuring is an offence under section 142 of the AML/CTF Act regardless of whether the underlying funds are from legitimate sources. Suspicion of structuring — not confirmation — triggers the SMR obligation.

Singapore — MAS

Under the Corruption, Drug Trafficking and Other Serious Crimes (Confiscation of Benefits) Act and MAS Notice SFA04-N02/CMS-N02 and related notices:

• Cash Transaction Report (CTR): Required for cash transactions of SGD 20,000 or more, or equivalent in foreign currency.
• Suspicious Transaction Report (STR): Must be filed with the Suspicious Transaction Reporting Office (STRO) within 1 business day of the institution's knowledge or suspicion.

Singapore's 1 business day STR deadline is among the strictest in the region.

Malaysia — BNM

Under the Anti-Money Laundering, Anti-Terrorism Financing and Proceeds of Unlawful Activities Act 2001 (AMLATFPUAA), regulated by Bank Negara Malaysia:

• Cash Threshold Report (CTR): Required for cash transactions of MYR 25,000 or more, or equivalent in foreign currency.
• Suspicious Transaction Report (STR): Must be submitted to the Financial Intelligence and Enforcement Department (FIED) within 3 working days of the institution forming a suspicion.

Philippines — BSP / AMLC

Under the Anti-Money Laundering Act of 2001 (Republic Act 9160) as amended, and rules issued by the Bangko Sentral ng Pilipinas (BSP) and the Anti-Money Laundering Council (AMLC):

• Covered Transaction Report (CTR): Required for single-day cash transactions totalling PHP 500,000 or more.
• Suspicious Transaction Report (STR): Must be filed with the AMLC within 5 business days of the transaction being deemed suspicious.

In all four jurisdictions, a failure to file — even where the transaction later proves legitimate — carries significant regulatory and criminal liability for the reporting institution.

Why Rule-Based Transaction Monitoring Misses Structuring

Traditional transaction monitoring systems work by evaluating individual transactions against a set of rules: flag any cash deposit over a threshold; flag any transaction to a high-risk jurisdiction; flag any customer who exceeds a monthly cash limit.

Structuring is engineered to defeat exactly this type of detection. Each individual transaction passes every rule. No single deposit exceeds the threshold. No single account exhibits abnormal volume. The problem only exists in the aggregate — across multiple transactions, multiple accounts, and an extended time window.

A rule that flags AUD 10,000+ deposits will not flag three AUD 9,500 deposits. A rule that flags high transaction frequency on a single account will not flag ten accounts each making one deposit per week.

For a broader explanation of how transaction monitoring systems work and what they are designed to catch, read our What is Transaction Monitoring blog.

The result is that structuring and smurfing schemes can run for months without generating a single alert, even in banks with fully implemented transaction monitoring programmes. The rules are working exactly as configured. That is the problem.

How Machine Learning-Based Systems Detect Structuring Patterns

The detection challenge is a data aggregation problem, and machine learning systems are better suited to it than rule-based engines for three specific reasons.

Velocity analysis across accounts and time

ML systems can calculate velocity — the rate of sub-threshold deposits — across a population of accounts simultaneously, and flag when a cluster of accounts shows a correlated spike. A rule fires when one account crosses a threshold. A velocity model fires when 12 accounts in the same network collectively accumulate AUD 95,000 across six weeks in increments designed to avoid individual-account triggers.

Network graph analysis

By mapping relationships between accounts — shared addresses, shared phone numbers, overlapping transaction counterparties — graph-based models identify structuring networks that appear unconnected at the individual account level. The smurfing structure that looks like 10 ordinary retail customers becomes a visible ring when the relationship layer is added.

Temporal pattern detection

Structuring schemes operate on a schedule. Deposits cluster on specific days of the week, at specific times, in specific amounts. ML models trained on transaction sequences can identify these temporal signatures and surface accounts that match them, even when the amounts are individually unremarkable.

The practical consequence is a material reduction in both false negatives (missed schemes) and false positives (unnecessary alerts). Rules generate noise. Pattern models generate signal.

If your institution is evaluating whether its current transaction monitoring system can detect structuring at the pattern level rather than the transaction level, the Transaction Monitoring Software Buyer's Guide covers the evaluation framework — including the specific questions to ask vendors about multi-account aggregation and network analysis capabilities.

The compliance team reviewing 400 alerts each morning cannot manually reconstruct an 11-week deposit pattern across 14 accounts. That is not an attention problem. It is a systems problem. Structuring detection requires systems built for pattern-level analysis, regulatory obligations that are jurisdiction-specific and time-bound, and an alert triage process that distinguishes genuine red flags from rule-based noise.

The technology to close that gap exists. The question is whether the system currently in place is designed to find it.

Australia’s financial system is changing fast, and a new class of AML and fraud prevention software vendors is defining what strong compliance looks like today.

Introduction
‍

Two AUSTRAC enforcement actions in three years — Commonwealth Bank's AUD 700 million settlement in 2018 and Westpac's AUD 1.3 billion settlement in 2021 — were both linked directly to failures in transaction monitoring and fraud detection software. Not the absence of a system. The failure of one already in place.

That context matters when Australian institutions are comparing AML and fraud prevention software. The decision is not which vendor has the best demo. It is which system will still be performing correctly when AUSTRAC examines it.

This guide covers the top vendors with genuine influence in Australia's AML and fraud prevention market, the five evaluation criteria that distinguish serious systems from adequate ones, and the questions to ask before committing to any platform. The list reflects deployment footprint and regulatory track record in Australia — not marketing spend.

Why Choosing the Right AML Vendor Matters More Than Ever
‍

Before diving into the vendors, it is worth understanding why Australian institutions are updating AML systems at an accelerating pace.

1. The rise of real time payments

NPP has collapsed the detection window from hours to seconds. AML technology must keep up.

2. Scam driven money laundering

Victims often become unwitting mules. This has created AML blind spots.

3. Increasing AUSTRAC expectations

AUSTRAC now evaluates systems on clarity, timeliness, explainability, and operational consistency.

4. APRA’s CPS 230 requirements

Banks must demonstrate resilience, vendor governance, and continuity across critical systems.

5. Cost and fatigue from false positives

AML teams are under pressure to work faster and smarter without expanding headcount.

The vendors below are shaping how Australian institutions respond to these pressures.

Top AML and Fraud Prevention Software Vendors in Australia
‍

1. Tookitaki

FinCense is Tookitaki's end-to-end AML and fraud prevention platform, built specifically for financial institutions in APAC. It combines transaction monitoring, fraud detection, screening, and case management within a single system — covering over 50 financial crime scenarios including account takeover, mule account detection, APP scams, trade-based money laundering, and real-time NPP-specific fraud patterns.

AUSTRAC alignment

FinCense is pre-configured with AUSTRAC-specific typologies, produces alert documentation in the format AUSTRAC examiners review, and supports direct generation of Threshold Transaction Reports (TTRs) and Suspicious Matter Reports (SMRs). Alert thresholds are calibrated to each institution's customer risk assessment — not applied from generic defaults — which directly addresses the calibration deficiencies that featured in AUSTRAC's 2018 and 2021 enforcement actions.

Real-time NPP processing

FinCense evaluates transactions pre-settlement, before NPP payments are confirmed irrevocable. This is a specific requirement for Australian institutions that batch-processing legacy systems cannot meet. Detection runs at the point of transaction initiation, not in end-of-day sweeps.

Federated learning and the AFC Ecosystem

FinCense's detection models are trained using federated learning across Tookitaki's AFC Ecosystem — a network of financial institutions that share anonymised typology intelligence without exchanging raw customer data. This means detection models reflect cross-institution fraud patterns, including coordinated mule account activity that moves between banks. Single-institution training data cannot surface these patterns.

False positive reduction

In production deployments, FinCense has reduced false positive rates by up to 50% compared to legacy rule-based systems. For a compliance team managing 400 alerts per day, that translates to approximately 200 fewer dead-end investigations — freeing analyst capacity for genuine risk signals.

Explainable alerts

Every FinCense alert includes a traceable rationale: the specific rule or model output, the customer history data points considered, and the risk factors that triggered the flag. This explainability supports both analyst decision quality and AUSTRAC audit documentation requirements.

Scalability

FinCense is deployed across institution sizes — from major banks to regional credit unions and PSA-licensed payment institutions. The platform scales to high transaction volumes without architecture changes, and implementation timelines are defined contractually rather than estimated.

Book a demo to see FinCense running against Australian fraud and AML scenarios.

For a detailed evaluation framework — including the 7 questions to ask any AML vendor before you sign — see our Transaction Monitoring Software Buyer's Guide.

2. NICE Actimize

NICE Actimize is a financial crime compliance suite from NICE Systems covering transaction monitoring, fraud detection, and sanctions screening. It is primarily deployed at large global financial institutions and has a long operational track record in the enterprise market.

3. SAS Anti-Money Laundering

SAS Anti-Money Laundering is part of SAS Institute's risk and compliance portfolio. It is an analytics-driven detection platform suited to institutions with established data science capabilities and high data maturity requirements.

4. SymphonyAI NetReveal

SymphonyAI's NetReveal is a financial crime management platform that blends established compliance protocols with advanced AI to detect fraud and money laundering. Originally acquired from BAE Systems, it now forms part of the Sensa-NetReveal Suite, which unifies traditional rules-based systems with cutting-edge predictive and generative AI.

5. Napier AI

Napier AI is a London-based financial technology company that provides a cloud-native, AI-enhanced platform for anti-money laundering (AML) and financial crime compliance. Founded in 2015, it is known for its "NextGen" approach, combining traditional rule-based systems with machine learning to reduce false positives and automate complex investigations.

6. LexisNexis Risk Solutions

LexisNexis Risk Solutions is a global data and analytics giant that provides risk intelligence across a massive range of industries, from banking and insurance to healthcare and law enforcement.

7. Quantexa

Quantexa is a London-based AI and data analytics leader specializing in Decision Intelligence (DI). Founded in 2016, the company focuses on "connecting the dots" between siloed data sources to reveal hidden relationships and risks.

What This Vendor Landscape Tells Us About Australia’s AML Market
‍

After reviewing the top vendors, three patterns become clear.

Pattern 1: Banks want intelligence, not just alerts

Vendors with strong behavioural analytics and explainability capabilities are gaining the most traction. Australian institutions want systems that detect real risk, not systems that produce endless noise.

Pattern 2: Case management is becoming a differentiator

Detection matters, but investigation experience matters more. Vendors offering advanced case management, automated enrichment, and clear narratives stand out.

Pattern 3: Mid market vendors are growing as the ecosystem expands

Australia’s regulated population includes more than major banks. Payment companies, remitters, foreign subsidiaries, and fintechs require fit for purpose AML systems. This has boosted adoption of modern cloud native vendors.

How to Choose the Right AML Vendor
‍

Buying AML and fraud prevention software is not about selecting the biggest vendor or the one with the most features. It involves evaluating five critical dimensions.

1. Fit for the institution’s size and data maturity

A community bank has different needs from a global institution.

2. Localisation to Australian typologies

NPP patterns, scam victim indicators, and local naming conventions matter.

3. Explainability and auditability

Regulators expect clarity and traceability.

4. Real time performance

Instant payments require instant detection.

5. Operational efficiency

Teams must handle more alerts with the same headcount.

Conclusion
‍

Australia’s AML and fraud landscape is entering a new era.
‍
The vendors shaping this space are those that combine intelligence, speed, explainability, and strong operational frameworks.

The top vendors highlighted here represent the platforms that are meaningfully influencing Australian AML and fraud landscape. From enterprise platforms like NICE Actimize and SAS to fast moving AI driven systems like Tookitaki and Napier, the market is more dynamic than ever.

Choosing the right vendor is no longer a technology decision.
It is a strategic decision that affects customer trust, regulatory confidence, operational resilience, and long term financial crime capability.

The institutions that choose thoughtfully will be best positioned to navigate an increasingly complex risk environment.

Choosing a transaction monitoring solution in Australia is a different decision than it is anywhere else in the world — not because the technology is different, but because the regulatory and payment infrastructure context is.

AUSTRAC has one of the most active enforcement programmes of any financial intelligence unit globally. The New Payments Platform (NPP) makes irrevocable real-time transfers the default for domestic payments. And Australia's AML/CTF framework is mid-way through its most significant legislative reform in fifteen years, with Tranche 2 expanding obligations to lawyers, accountants, and real estate agents.

For compliance teams at Australian reporting entities, this means a transaction monitoring solution needs to do more than pass a vendor demonstration. It needs to perform under AUSTRAC examination and keep pace with payment infrastructure that moves faster than most legacy monitoring systems were designed for.

This guide covers what AUSTRAC actually requires, the criteria that matter most in the Australian market, and the questions to ask before committing to a solution.

What AUSTRAC Requires from Transaction Monitoring

The AML/CTF Act requires all reporting entities to implement and maintain an AML/CTF programme that includes ongoing customer due diligence and transaction monitoring. The specific monitoring obligations sit in Chapter 16 of the AML/CTF Rules.

Three points from Chapter 16 matter before any vendor evaluation begins:

Risk-based calibration is mandatory. Monitoring thresholds must reflect the institution's specific customer risk assessment — not vendor defaults. A retail bank, a remittance provider, and a cryptocurrency exchange each need monitoring calibrated to their own customer profile. AUSTRAC does not prescribe specific thresholds; it assesses whether the thresholds in place are appropriate for the risk present.

Ongoing monitoring is a continuous obligation. AUSTRAC expects transaction monitoring to be a live function, not a periodic review. The language in Rule 16 about real-time vigilance is not advisory — it reflects examination expectations.

The system must support regulatory reporting. Threshold Transaction Reports (TTRs) over AUD 10,000 and Suspicious Matter Reports (SMRs) must be filed within regulated timeframes. A monitoring system that cannot generate AUSTRAC-ready reports — or that requires significant manual handling to produce them — creates compliance risk at the reporting stage even when the detection stage works correctly.

The enforcement record illustrates what happens when monitoring falls short. The Commonwealth Bank of Australia's AUD 700 million AUSTRAC settlement in 2018 and Westpac's AUD 1.3 billion settlement in 2021 both named transaction monitoring failures as direct causes — not the absence of monitoring systems, but systems that failed to detect what they were required to detect. Both cases involved institutions with significant compliance investment already in place.

The NPP Factor

The New Payments Platform reshaped monitoring requirements for Australian institutions in a way that most global vendor comparisons do not account for.

Before NPP, Australia's payment infrastructure gave compliance teams a window between transaction initiation and settlement — a clearing delay during which a flagged transaction could be investigated before funds moved irrevocably. NPP eliminated that window. Domestic transfers now settle in seconds.

Batch-processing monitoring systems — even those with short batch intervals — cannot catch NPP fraud or structuring activity before settlement. The only viable approach is pre-settlement evaluation: risk assessment at the point of transaction initiation, before the payment is confirmed.

When evaluating vendors, ask specifically: at what point in the NPP payment lifecycle does your system evaluate the transaction? Vendors frequently describe their systems as "real-time" when they mean near-real-time or fast-batch. That distinction matters both for fraud loss prevention and for AUSTRAC examination.

6 Criteria for Evaluating Transaction Monitoring Solutions in Australia

1. Pre-settlement processing on NPP

The technical requirement above, stated as a discrete evaluation criterion. Ask for a live demonstration using NPP transaction scenarios, not hypothetical ones.

2. Alert quality over alert volume

High alert volume is not a sign of effective monitoring — it is often a sign of poorly calibrated thresholds. A system generating 600 alerts per day at a 96% false positive rate means approximately 576 dead-end investigations. That is not compliance; it is operational noise that crowds out genuine risk signals.

Ask for the vendor's false positive rate in production at a comparable Australian institution. A well-calibrated AI-augmented system should be below 85% in production. If the vendor cannot provide production data from a comparable client, that is itself informative.

3. AUSTRAC typology coverage

Australia has specific financial crime patterns that global rule libraries do not always cover — cross-border cash couriering, mule account networks across retail banking, and real estate-linked layering using NPP for settlement. These typologies are documented in AUSTRAC's annual financial intelligence assessments and should be represented in any system deployed for an Australian institution.

Ask to see the vendor's AUSTRAC-specific typology library and when it was last updated. Ask how the vendor tracks and incorporates new AUSTRAC guidance.

4. Explainable alert logic

Every AUSTRAC examination includes review of alert documentation. For each sampled alert, examiners expect to see: what triggered it, who reviewed it, the analyst's written rationale, and the disposition decision. A monitoring system built on opaque models — where alerts are generated but the logic is not traceable — makes this documentation impossible to produce correctly.

Explainability also improves investigation quality. An analyst who understands why an alert was raised makes a better disposition decision than one who cannot reconstruct the reasoning.

5. Calibration without constant vendor involvement

AUSTRAC requires monitoring thresholds to reflect the institution's current customer risk profile. Customer profiles change: books grow, customer mix shifts, new products are launched. A monitoring system that requires a vendor engagement to update detection scenarios or adjust thresholds will always lag behind the institution's actual risk position.

Ask specifically: can your compliance team modify thresholds, create new scenarios, and adjust rule weightings independently? What is the governance process for documenting calibration changes for AUSTRAC audit purposes?

6. Integration with existing case management

Transaction monitoring does not exist in isolation. Alerts feed into case management, case management informs SMR decisions, and SMR decisions must be filed with AUSTRAC within regulated timeframes. A monitoring solution that requires manual data transfer between systems at any of these stages creates delay, error risk, and audit trail gaps.

Ask for the vendor's standard integration points and reference implementations with Australian case management platforms.

Questions to Ask Before Committing

Most vendor sales processes focus on features. These questions get at operational and regulatory reality:

Do you have current AUSTRAC-supervised clients? Ask for references — not case studies. Speak to compliance teams at comparable institutions running the system in production.

How did your system handle the NPP real-time payment requirement when it was introduced? A vendor's response to an infrastructure change already in the past tells you more about adaptability than any forward-looking roadmap.

What is your typical time from contract to production-ready performance? Not go-live — production-ready. The gap between those two dates is where most implementation budgets fail.

What does your model retraining schedule look like? Transaction patterns change. A model trained on 2023 data that has not been retrained will underperform against current fraud and laundering patterns.

How do you handle Tranche 2 obligations for our institution? For institutions with subsidiary or affiliated entities in Tranche 2 sectors, the monitoring solution needs to be able to extend coverage without a separate implementation.

Common Mistakes in Vendor Selection

Three patterns appear consistently in post-implementation reviews of Australian institutions that struggled with their monitoring solution:

Selecting on cost rather than calibration. The cheapest system at procurement often becomes the most expensive when AUSTRAC examination findings require remediation. Remediation costs — additional vendor work, internal team time, reputational risk management — typically exceed the original licence cost difference many times over.

Underestimating integration complexity. A system that performs well in isolation but requires significant custom integration with the institution's core banking platform and case management tool will consistently underperform its demonstration capabilities. Ask for the implementation architecture documentation before signing, not after.

Treating go-live as done. Transaction monitoring requires ongoing calibration. Banks that deploy a system and then do not actively tune it — adjusting thresholds, adding new typologies, reviewing alert quality — see performance degrade within 12–18 months as their customer profile evolves away from the profile the system was originally calibrated for.

How Tookitaki's FinCense Works in the Australian Market

FinCense is used by financial institutions across APAC including Australia, Singapore, Malaysia, and the Philippines. In Australia specifically, the platform is configured with AUSTRAC-aligned typologies, supports TTR and SMR reporting formats, and processes transactions pre-settlement for NPP compatibility.

The federated learning architecture allows FinCense models to incorporate typology patterns from across the client network without sharing raw transaction data — which means Australian institutions benefit from detection intelligence learned from cross-institution fraud patterns, including coordinated mule account activity that moves between banks.

In production, FinCense has reduced false positive rates by up to 50% compared to legacy rule-based systems. For a team managing 400 daily alerts, that translates to approximately 200 fewer dead-end investigations per day.

Next Steps

If your institution is evaluating transaction monitoring solutions for 2026, three resources will help structure the process:

• AUSTRAC Transaction Monitoring Requirements — detailed breakdown of Chapter 16 obligations and what AUSTRAC examines in practice
• Transaction Monitoring Software Buyer's Guide — the 7 questions to ask any vendor before you sign
• What Is Transaction Monitoring? — the complete technical and regulatory overview

Or talk to Tookitaki's team directly to discuss your institution's specific requirements.