Adaptive security for fintech
Our Mission
We protect automated decisions against evolving online fraud. Resistant AI connects the dots to provide a new layer of trust and performance for our clients’ systems.
How We Got Started – Security DNA
Our team of research and industry experts, seasoned in AI security and machine learning, has embarked on its next challenge after our previous startup company, Cognitive Security, where we applied machine learning to computer network security, was acquired by Cisco Systems. The company has since grown to protect more than 25 million users worldwide.
Our Customers
- Banks
- Non-banking financial institutions
- E-commerce
- Insurance companies
Our Investors
- Index Ventures
- Credo Ventures
- Seedcamp
Our Products
Resistant Documents
Document Forgery Detection
Resistant Documents protects automated processes that rely on unauthenticated documents received from third parties in pdf, jpeg and other picture formats. Typical documents include invoices, payroll slips, bank statements, KYC documents, etc. Resistant Documents detects forgeries and modifications, that are related to identity changes, fraudulent modification of account numbers and transaction details.
Key Benefits
- Automatically detect fraud
- Reduce fraud losses due to forged or manipulated documents
- Relieve your operational teams of manual document verification
- Automatically flag and annotate problematic cases for your fraud team
- Easy and fast integration via API
Our Customers
- Online lending
- Consumer and small business
- Car financing
- Factoring
- Insurance
Product Details
Machine learning models
Based on a robust collection of machine learning models that represent the behavior of document issuers (such as banks or payroll companies). The system also models financial software, software libraries and even individual devices such as scanners or mobile phones. The system continuously learns and improves its detection performance with each document it assesses, both legitimate or malicious.
No black box
Each document flagged by the system is justified by human-readable risk indicators presented to the fraud analyst.
Simple to use, fast to integrate
The service is delivered via a simple REST API and returns results in mere seconds. Also available for UiPath.
Improved performance
Advanced fraud detection measures by automating risk assessments, allowing operational and fraud investigation teams to concentrate on high-risk cases.
Resistant Transactions
Transaction Association Engine
Resistant Transactions protects online credit scoring, fraud detection and Anti-Money-Laundering (AML) systems against manipulation and circumvention by advanced and organised fraudsters. It also protects the fraud teams against false alarms and wrongful rejections caused by sub-optimal training, large rule-sets or other reasons.
High level of protection automation is critical in online and omnichannel settings. While online business is fast, online fraud can be equally far more damaging due to high transaction throughput and missing protections that manual processing provides. Our system has been designed to prevent organised attacks in the early stages of their activity.
Key Benefits
- Early-stage detection of organised fraud and prevention of major losses
- Reduction of operational costs caused by false alarms
- Fraud operations can be focused on high-value, high-risk operations early in the decision process
- Clearly labeled fraud data separated from default data improves the underlying credit scoring process
Our Customers
- Banks
- Non-banking financial institutions
- E-commerce
Product Details
Security Monitoring
Today’s fraudsters have learned to live in a world shaped by automated decisions. Knowingly or just by intuition, fraudsters can identify the vulnerabilities in automated decisions and exploit them at scale. Resistant Transactions discovers early-stage symptoms of such attacks and provides an early warning before a substantial loss occurs.
Connecting the dots
Our product provides additional protection by identifying and leveraging similarities between seemingly unrelated transactions. We use machine learning techniques to distinguish mere coincidences from unusual groups of related transactions.
In-Line Transaction Blocking
Fraud detection models designed in Security Monitoring can be used in-line to prevent similar incidents from occurring in the future. This not only protects our customers from financial losses, but also prevents the fraudsters from gathering information about vulnerabilities of the underlying system.
False Positive Reduction
False alerts result in increased analyst workload and prevent the analysts from concentrating on high-risk fraud. Each false rejection sends a valuable customer straight to your competition. Resistant Transactions can filter-out false alerts and prioritise high risk alerts based on internal self-learning mechanisms. Our system prioritises key alerts from the underlying system and packages them together for actionable analyses based on full transaction context.
Rule-Based System Optimisation
A specific case of false positives reduction is the AML domain, where decisions are driven by rule-sets received from different sources. Manual management and lack of optimisation of complex rule-based systems can lead to operational inefficiencies or regulatory failures. Resistant Transactions help the AML teams efficiently prioritise their efforts and deliver high-quality inspections with reasonable effort.
Delivery
In-line transaction blocking is delivered via a high-throughput in-line rest API and periodic batch uploads used for Security Monitoring.
FAQ
What is our technology?
We constantly monitor the protected system. This includes the analysis of inputs and outputs, as well as the verification of the system's security and robustness against various types of frauds and AI attacks. Using the same methodology, we also shield traditional statistical models and rule-based systems from AI-enabled attacks or fraudulent manipulation. We use custom advanced statistical approaches, based on machine learning techniques, that enable us to immediately detect emerging fraud behaviour.
What are key benefits for our customers?
Forgery detection. Prevent fraudsters from manipulating your business decisions by submission of forged or altered documents.
False positive reduction. Improve the effectiveness of risk decision systems (fraud detection, credit risk, Anti-Money-Laundering (AML), payment) and reduce operational costs.
Advanced and organised fraud detection. Identify and prevent organised fraud, technologically advanced fraud or internal fraud and mitigate associated losses. Resistant AI uses advanced statistical and machine learning techniques for early identification of groups of transactions leading to major fraud losses.
Where can we meet?
We are based in Prague and Brussels and cater to the needs of customers worldwide.
References
Michal Krocil Chief Risk Officer, Twisto payments
Resistant AI’s oversight solution is able to dramatically reduce false positives and detect advanced fraud and manipulation at the same time. Under the protection of Resistant AI, our Nikita engine can fully focus on credit risk assessment excellence.
Michal Tresner CEO, ThreatMark – leading provider of biometric authentication for the banking sector
Our experience with Resistant AI has been eye-opening. Our detection engine moved a generation ahead in terms of security resistance and statistical robustness. As a by-product, the software robustness has increased as well.
Alexandra Belkovova Head of Operations, Payment institution Roger a.s.
Resistant AI prevents manipulation of invoices submitted to our marketplace. It allows our investors to trade in security and saves my team a huge amount of manual work.
Contact
How can we help?
Email: sales@resistant.ai
Phone: +420 704 470 207
Resistant AI is based in Prague and Brussels.
Investors
