Features

One platform to control your fraud risk

YARDA trains AI models on your own transactions, with pseudonymised identifiers, and improves them from the cases your teams label. Detection adapts to your business, not to generic rules.

Models that grow in stages

Two AI models, put into service one after the other. First, an anomaly detection model trained on your transactions, with no labels required. Next, a supervised classifier trained on the cases your teams label with your own fraud taxonomy. A graph neural network is being researched; it plays no part in scoring.

Pseudonymised data, privacy by design

YARDA does not need your customers’ names or addresses. The identifiers you send, phone numbers included, are replaced with pseudonyms as soon as they arrive, before anything is stored, with a key specific to your institution. The data you send is still personal data under the regulations, and we process it as your processor, as described in our privacy policy.

Analysis and compliance console

Monitor transactions and their scores, work through the alert queue, label each case with your own fraud taxonomy so the models improve, and follow how your risk exposure changes.

Explainable alerts

Every alert comes with the factors that triggered it and the contribution of each model. Your analysts understand why a transaction was flagged and can justify their decisions to your auditors and the authorities.

An isolated space for each operator

Each operator has its own models, its own settings and data isolated from everyone else’s. Adjustable alert thresholds, a fraud taxonomy suited to your business, and transactions kept in their original currency.

Retraining on your labels

The cases your teams label are used to retrain your models. A new supervised model first runs in observation: it only goes live if it outperforms the current one.

A free diagnostic, a pilot with success criteria, then a subscription in CFA francs based on your volume.

Pricing

See YARDA on your own transactions

Tell us who you are: we will get back to you to arrange a demo, then a free diagnostic on your history. No development, no commitment.

What happens next

  1. A demo of the console, on fictitious data
  2. A free diagnostic on three to six months of your history
  3. A pilot with success criteria agreed together