Graphen

Voice and text assistants that answer company questions

Graphen says it helps enterprises build voice-based dialogue apps for smartphones. These are industrial AI assistants that can handle questions about a company, whether they arrive by voice or by text, and they are presented as a way for enterprises to scale their customer service without adding staff at the same rate. The assistants answer company-related questions and are designed for the kind of high volume that a contact center would otherwise have to handle by hand.

Real-time fraud detection for banks and payments

Much of the firm's work is fraud detection. Its real-time transaction system brings together customer data and transaction information to identify the risk of fraud in customer transactions. A separate insurance fraud product uses graph analysis to uncover hidden relationships between parties, then applies machine learning to behavioral models built from known suspicious patterns. A financial agent fraud system combines artificial intelligence with a graph database to detect potential fraud by the agents who work for financial institutions, which the firm says protects banks from the inside.

Anti-money laundering and due diligence for merchants

The anti-money laundering system imports behavior models to detect laundering patterns, and it uses a machine learning module to keep those models optimized. The know your customer product, which the firm calls smart due diligence, is intended to reduce the workload that merchant due diligence and payment operations place on banks while strengthening their know your customer checks. It applies a knowledge graph and Bayesian network probabilities to monitor suspicious activity and merchants. The firm says the same approach suits any industry that wants deeper customer insight, including online marketing and e-commerce.

Monitoring company networks for internal threats

The security line uses artificial intelligence and graph analysis to scan a company's information network and learn the normal behavior of each terminal. When an anomaly and its likely motive are identified, the system is designed to predict what abnormal behavior may come next. The firm lists long-term threats, internal threats such as leakage and sabotage, and IoT security as areas where this anomaly detection applies. Its business monitoring system, run from a central control center, shows an organization's operations in real time, including online and mobile activity and abnormal conditions on computers and other equipment.

Wealth advice, energy forecasting and other specialist applications

Graphen also describes an AI personal financial advisor that supports wealth management and investment. In energy, its Ardi platform is used for electric field and power system anomaly analysis, and for forecasting distributed energy sources and loads. The firm says it has helped one of the largest energy providers in Asia make hourly 24-hour predictions of solar power output at each station in Taiwan. Its health line offers a whole-genome analytics system that reports risk across more than 350 diseases, and a drug design service that predicts protein structures from genes. Its automotive work includes a car diagnosis system that suggests fixes, which the firm reports reached close to 99% accuracy. The firm also describes its work on AI robots that understand their environment through language and computer vision.