Xanthos is a professional pseudo-transaction generator: it produces realistic, fully synthetic transaction streams against which fraud and AML systems can be evaluated. Balius is its monitor: a measurement tool that watches how systems respond to those streams. Together they give risk and compliance teams an honest answer to the question "does our defence actually work?"
Both tools are available to market participants by registration. We ask for a modest, optional contribution — and every contribution we receive funds financial-literacy education for children.
Recorded historic data is convenient but limited: it cannot include the adversarial patterns the team needs to test against, and it brings real-world privacy and licence baggage into the test environment. Xanthos solves both problems by generating realistic, fully synthetic streams.
Velocity, amount, counterparty fan-out, currency mix, time-of-day cadence — the statistical shape of the synthetic stream matches what production systems actually receive, so the system under test cannot tell it from real traffic in shape.
Specific typologies — structuring, smurfing, mule-network fan-out, bust-out fraud, layering chains — can be injected at controlled prevalence levels. The team running the test knows exactly where the needles are and can score the haystack accordingly.
No real customers, no real accounts, no real identities. Streams can be reproduced, archived and shared between teams without privacy, GDPR or banking-secrecy concerns — they are simply not personal data.
Streams are seeded and deterministic. The same seed produces the same stream every time, so test results are comparable across runs, across versions of the system under test, and across teams.
A test is only as good as the metrics it captures. Balius watches how the system under test responds to the Xanthos stream and produces evidence-grade artefacts the test team can trust.
Because Xanthos knows where the adversarial needles are, Balius can give a clean confusion matrix — the kind of measurement that is genuinely difficult to obtain in production environments where ground truth is mostly unknown.
Per-event timing, broken down by event type and load level. Tail-latency behaviour under adversarial volume — the part that matters in production — is captured separately from average behaviour.
How accuracy and latency degrade as throughput rises towards realistic peak conditions. Necessary input for the operational and capacity-planning decisions teams need to defend in front of a steering committee.
Every Balius run produces a signed, reproducible report — including the Xanthos seed, the system version under test, the metrics, and the underlying per-event traces. The kind of artefact internal audit and external auditors actually accept as evidence.
The deepest defence against being exploited financially is understanding how finance works. Predatory lending, fraud, scams, and the financial routines that underpin organised crime are easiest to deploy against people who have not been prepared for them. Children are particularly exposed.
Market participants — risk and compliance teams, university research groups, partner labs — register for Xanthos & Balius access. We ask for a modest, fully optional contribution per access tier. The contribution is the price of a self-respecting professional tool, not the price of a vendor product.
Every contribution received through Xanthos & Balius is dedicated to financial-literacy programmes for children — not blended into Silbad's general operating budget, not used to pay anyone's salary, and not retained as profit. The dedication is operational, not aspirational.
Once enough has been received to fund a programme, we publish — concretely and transparently — what was funded, where, and what it produced. The first report will appear here on the silbad.eu site once we have something tangible to report. We commit not to publish it before we do.
This is the same mission, expressed from a different end of the financial-crime problem. The four institutional products — Bucephalus, Marengo, Babieca, Chetak — defend against IFF / CFT / SFMA at the institutional level. Xanthos & Balius add a defence at the source: making the next generation harder to exploit.
Xanthos and Balius are professional tools for professional testing. Access is open to legitimate market participants on registration — not to anonymous public download — because we need to know who is testing what, and because the contribution flow has to be auditable.
Banks, payment institutions, fintechs and other regulated entities testing their own AML / fraud / sanctions stacks. Standard institutional access tier; contribution scaled to institution size.
University labs and independent research groups studying financial-crime detection, federated learning, or anomaly detection on transactional data. Reduced-contribution tier, on confirmation of academic affiliation.
Independent third-party testing organisations operating commercially in the financial-crime defence space. Standard institutional access tier, with additional reproducibility commitments on both sides.
Regulators and supervisors using the tools to evaluate vendor offerings or to define test methodology. Access by direct arrangement; contribution waived where the authority's procurement frame requires it.
