Does the platform replace our core tax system?

No. It is the analysis and planning layer alongside the systems that register taxpayers, receive returns and record payments. Data is exported from those systems as structured files or through an API, and the mappings are configured during deployment.

Does it assess tax, impose penalties or stop refunds by itself?

No. It flags, ranks and recommends. Assessments, penalties, refund holds and every other action are decided by officials under the authority’s own legislation and delegations, and each decision is recorded with the person who made it.

How are the revenue estimates and recommendations made?

Revenue at risk is estimated from the gap between what was declared and what the data indicates. Expected yield for each intervention is learned from the authority’s own past results and recalibrated as new outcomes are recorded. Each estimate shows the data behind it.

Where is taxpayer data held?

In-country, on infrastructure the authority controls: on premises, in a government cloud or in a hybrid arrangement, consistent with tax secrecy rules such as Chapter 6 of South Africa’s Tax Administration Act. Access follows role and need-to-know, sensitive identifiers can be tokenized in the PII vault, and every access and decision is written to a tamper-evident audit trail.

Does it cover customs as well as domestic taxes?

Yes. Customs declarations can be loaded alongside domestic tax data, so the same importer is seen across both, and declared values are compared with comparable consignments of the same goods.

How does it fit how revenue authorities manage compliance risk?

It follows the cycle the OECD Forum on Tax Administration set out for compliance risk management: identify the risks, assess and prioritise them, choose a treatment for each and evaluate the outcome. The platform does the identifying, ranking and measuring at scale, and officials choose the treatments.

How does it work with South African tax law?

Workflows are configured to the authority’s legislation and delegations. In South Africa, for example, the third-party returns the Commissioner can require under section 26 of the Tax Administration Act feed the data matching, and a refund flagged as high risk can be held for verification under section 190(2) by an official’s decision. Customs falls under the Customs and Excise Act 91 of 1964, not the Tax Administration Act, and is configured separately.

FinCrimes

Find where revenue leaks, and the fastest way to the revenue target

A revenue authority has a target to meet and a fixed number of people to meet it with. Revenue leaks where businesses trade without registering, declare less than third-party data shows, claim refunds they are not owed or under-value imports. FinCrimes brings the authority’s own returns, registrations, payments and third-party data into one view of every taxpayer, uses AI to find and rank the leaks, and recommends which interventions to run first to close the gap to target. People decide, and every recommendation shows the evidence behind it.

See how it works

Part of Dark Pools FinCrimes

Capabilities

What revenue authorities gives your team

One view of every taxpayer

Registrations across tax types, returns, payments, customs declarations and third-party data resolve into one record per person and business, with the links between them: directors, addresses, bank accounts and the tax practitioners who file for them.

Entity resolution across income tax, VAT, employer and customs registrations
Third-party data matched to the taxpayer it describes: banks, employers, company and property registers
Related parties shown on a link chart, so a group of companies is seen as a group
Each fact keeps its source, so an auditor can see where it came from

AI leakage detection

Machine-learning models and explainable rules compare what each taxpayer declares with what the data shows and with comparable taxpayers in the same sector and size band.

Businesses trading without registering for the taxes they owe
Turnover or income declared below what third-party data indicates
Input claims, expenses and refund ratios far outside sector peers
Taxes withheld from employees or customers but not paid over
Imports declared well below comparable consignments of the same goods
Registered taxpayers who stop filing while their activity continues

Refund fraud and syndicate networks

Graph analytics exposes the networks that single-return reviews miss: newly registered vendors sharing bank accounts, addresses or directors, chains of invoices between related companies, and companies that close with tax owing while the same people open new ones.

Refund claims scored before payment, so high-risk refunds can be held for verification
Cycle detection across invoice chains between related suppliers
Shared attributes linking claimants that look unrelated one at a time
Each network opened as a case with the linked returns and evidence attached

Revenue at risk, ranked and explained

Every flag carries an estimate of the revenue involved and the reasons behind it, so the authority works the largest, best-evidenced leaks first and can defend each selection.

Estimated revenue at risk for each taxpayer, network and leak type
Each score shows the declarations, third-party records and peer comparisons that raised it
Versioned rules and models, with shadow scoring and backtesting before a change goes live
Selections that can be explained to the taxpayer, a tribunal or an oversight body

Strategies to reach the revenue target

Leadership sets the target by tax type, sector and region. The platform tracks collections against it, shows which leaks drive the gap, and recommends the mix of interventions expected to close it within the capacity the authority has.

Progress against the revenue target, with the gap broken down by tax type, sector and region
Options compared by expected yield and effort: letters, registration drives, desk reviews, audits, refund verification and customs inspections
A recommended plan that fits the auditors, inspectors and time available
Scenarios to test before deciding, such as moving capacity from one tax type to another

Learn what works

The result of every intervention is recorded against the leak it targeted: revenue raised, assessments upheld and whether the taxpayer complied afterwards. Those results recalibrate the estimates behind the next plan.

Yield and compliance recorded for each intervention and letter variant
Models retrained on outcomes, with a person approving each new version
Results reported by tax type, sector and region for leadership and oversight bodies

How it works

From data to decision

01ConnectBring in registrations, returns, payments, customs declarations and the third-party data the authority already receives.
02ResolveMatch every record to one view of each taxpayer and the people and companies around them.
03DetectAI and explainable rules flag leaks and estimate the revenue at risk in each.
04PlanRank the leaks and recommend the interventions that close the most of the gap to target with the capacity available.
05ActOfficials approve and run the letters, reviews, audits, refund holds and inspections, each as a tracked case.
06LearnRecord what each intervention yielded and feed the results into the next plan.

In the field

Where it is used

GovernmentA VAT refund syndicateNewly registered vendors claim large refunds on invoices from suppliers that turn out to be connected. Each claim looks reasonable on its own, so refunds are paid before anyone sees the pattern.Refunds held before the money leaves GovernmentTurnover that third-party data contradictsThousands of businesses declare turnover well below what bank, card and customs data show, but the gaps sit in separate systems and auditors can only reach a small share of them.Undeclared revenue found in data already held GovernmentA mid-year gap to the revenue targetHalfway through the fiscal year, collections trail the target and audit capacity is fixed. Leadership has to decide quickly where to put its people.A plan to close the gap, built on evidence GovernmentImports declared below their valueImporters declare consignments at values far below comparable goods from the same origin, and inspections are spread too thinly to catch them.Inspections aimed where the revenue is

FAQ

Questions buyers ask

Get started

Take the next step

See revenue authorities working on your own data, deployed on infrastructure you control.

Explore FinCrimes