Oscilar
Oscilar sells one platform for the risk decisions a bank or fintech makes about its customers: onboarding, fraud, anti money laundering compliance and credit underwriting. Risk teams build and backtest policies without code, by drag and drop or in plain language, and more than 150 data partners plug in through its marketplace. Twenty five AI agents in its Agent Hub handle screening, KYC and KYB review, first level fraud and AML review and SAR narrative drafting, and analysts confirm every recommendation. Credit models score balance, cash flow, spend power and payment behavior.
SoFi, Nuvei and Clara report measured results, Core Bank uses its agents for AML investigations at its sponsor bank business, and dLocal runs AML monitoring, sanctions screening and case management on it. Nacha lists Oscilar as a Preferred Partner. In October 2026 it released an AI assistant and an MCP server, and TransUnion launched two fraud scores first through Oscilar.
Capability Axes
Capability grades
15 of 15 axes rated · 11 graded A or B
Twenty five agents listed in Oscilar's Agent Hub do real work, from KYC and KYB review and QC review to fraud network detection and SAR narratives, beside a family of credit models and an AI assistant. Underneath is a no code decisioning engine where a risk team composes policies. Take the models away and that engine still runs rules, so the AI adds speed and reach to a product that would stand without it.
Every agent recommendation goes to an analyst, and Oscilar states that all agent outputs are human confirmed. Policies are backtested and can be A/B tested before they go live, and role based access control governs who acts. The AI assistant and MCP server run under each user's own role. Document extraction returns confidence scored fields, but the thresholds that route work to a person are not published.
Oscilar names its credit models: Balance Prediction, Cashflow Score, Spend Power and Pay Prediction. That gives a validator something specific to interrogate. Policies are backtested and A/B tested before release, and a test case creation agent ships with the product. Model documentation, error rates for those models and drift monitoring cadence are not published.
Named customers carry figures. Nuvei cut manual underwriting time by 50%. SoFi raised processing speed by more than 30% and reports 50% faster time to market. Clara made onboarding three times faster, with three to four times the throughput across more than 700,000 decisions. Dibsy reports 50% fewer manual reviews, and Fluz reports approvals up 20%. Aggregate claims such as 45% fewer false positives trace to no named deployment, and the named results carry no baseline period.
Contract terms bar Oscilar's outside AI providers from keeping customer data or training on it. Oscilar says nothing comparable about its own use. Whether one institution's data or analyst feedback improves models served to other customers, and what retention applies inside Oscilar, are not stated.
A subprocessor list names each party with its purpose and hosting location, which tells a bank privacy office which third parties can touch its data. The privacy statement, updated in April 2026, covers Oscilar only as controller, excludes customer data and gives no retention period.
SOC 2 Type II, ISO 27001 and PCI DSS reports are each available on request, and third party penetration testing runs at least yearly. Report dates, audit periods and incident notification practice are not published. No AI management certification such as ISO 42001 is listed.
Nacha welcomed Oscilar as a Preferred Partner for account validation, fraud monitoring, and risk and fraud prevention on December 11, 2025, and lists it on its own site. That admission comes from the body that runs the ACH network. Sanctions, PEP and adverse media screening agents and a SAR narrative agent map the product onto Bank Secrecy Act work.
Credit underwriting from bank transaction data falls under Equal Credit Opportunity Act rules, and those inputs track income volatility, employment sector and geography. Oscilar's AI FAQ states that models undergo bias and disparate impact testing before deployment, and its TransUnion fraud scores come with adverse action reason codes. No method, results or fair lending documentation stand behind the testing claim. A bank's fair lending team will want the test design, the populations tested and the outcome.
Real time performance monitoring lets an institution catch a bad rule and correct the system after it reaches applicants. Identity precision of 99.99 percent is claimed with no method or denominator, and no accuracy guarantee, remediation term or correction route for a declined borrower is published.
Oscilar's public subprocessor list names OpenAI and Anthropic for AI analytics, Chalk.ai, Databricks and Redis for feature stores, and IPQS for fraud intelligence. A bank's vendor review can see which model providers touch its data. The list carries no date.
More than 150 data partners connect through a published marketplace, covering identity, bank account and credit data. TransUnion released its synthetic fraud and credit washing scores first through Oscilar in October 2026. An MCP server, generally available since October 2026, lets a team work with the platform from its own AI tools. Workflows can be composed without code or generated from plain language.
Amazon Web Services hosting runs in the United States by default with European and other regions optional, and Google Cloud also defaults to the United States with other regions available. The privacy statement places servers in the United States and uses standard contractual clauses for transfers. No on premises or hybrid option is described.
Consolidation is the commercial argument, one system replacing separate fraud, compliance and credit tools, yet no price is published to weigh against the licenses it displaces. The pricing page leads to a demo request, and the terms say only that billing is by invoice in USD.
Banks, credit unions and sponsor banks each get dedicated material, and the customer list runs across fintechs and payment companies such as SoFi, Nuvei, dLocal and Conta Azul. Coverage spans onboarding, fraud and sanctions screening, underwriting, disputes and collections. Oscilar hired a general manager for Latin America in October 2026.
Product updates
Material product, regulatory, evidence and commercial changes at Oscilar, each verified against a live source and tagged to the capability axis it bears on. Funding rounds and awards are not product changes and are not logged.
Oscilar disclosed a self hosted tabular foundation model, a transformer pretrained on structured data, that scores fraud, AML, credit and device risk by reading each customer's own labeled history as context at decision time instead of retraining per customer XGBoost models. Oscilar says it runs under 100 milliseconds, updates its context in 5.8 seconds, keeps SHAP attributions and reason codes, can be distilled to a standard XGBoost model, and is already in production for managed fraud services customers, with fine tuned versions for ACH, AML, card deposit and device signals.
TransUnion launched Synthetic Fraud and Credit Washing Scores that carry adverse action reason codes, and named Oscilar's risk platform as the first to incorporate them. Within Oscilar, the scores and their reason codes run alongside credit decisioning in one workflow, so a lender can act on synthetic identity or credit washing risk at account opening and cite it on an adverse action notice alongside the credit reasons.
Oscilar made its MCP Server generally available, so Claude, Claude Code or any MCP compatible assistant can connect to an Oscilar environment, turn a written credit or fraud policy into a draft workflow, test it, and work cases. The assistant runs under the user's existing roles, permissions and audit trail, and any workflow change it makes is saved as a draft that someone on the team deploys. Oscilar AI, the in product assistant that explains decisions rule by rule, went generally available at the same time, and both are included at no additional cost.
Compared With
Most editorial comparisons pair two vendors the index assesses as direct competitors for the same buyer. Some pair vendors that are adjacent rather than rival, where the useful question is where one ends and the other begins. Each carries a verdict, the buyer conditions that favor each vendor, and a graded side by side.
Alternatives to Oscilar
Other products that do the same kind of work as Oscilar, and what each one does.
Sardine's platform reads the device and the behavior behind every session, then uses those signals for fraud prevention, anti money laundering compliance and credit underwriting.
Featurespace, a Visa solution since Visa completed its $946 million acquisition in December 2024, sells real time fraud and financial crime detection to banks, card processors, acquirers and credit unions.
Feedzai sells real time fraud, scam and financial crime prevention to banks, card networks, processors, acquirers and merchants.
TrustDecision, headquartered in Singapore with offices in Malaysia, Indonesia, the Philippines, the United Arab Emirates, Saudi Arabia, Mexico and Hong Kong, sells fraud, credit and identity risk technology to banks, consumer lenders, payment companies and digital businesses.
SentiLink sells identity fraud scores that US banks, credit unions and fintechs run at the moment of application.
Riskified decides in under a second whether to approve each online order, and pays the merchant back for any fraud chargeback on an order it approved.
Pricing
Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.
No pricing data has been verified for this vendor. Pricing information will be published here once confirmed through vendor disclosure or third-party estimation.