Oscilar Launches Agent Hub, the First Unified AI Agent Platform Spanning Fraud, Compliance, Credit, and Onboarding

• More than 100 financial institutions — including SoFi, MoneyGram, and Nuvei — trust Oscilar to process tens of billions of risk decisions per year
• SoFi cut time-to-market by 50%, Nuvei reduced alert review time by 50%, and Clara achieved a 4x increase in underwriting throughput on Oscilar's agentic risk platform
• Agent Hub coordinates more than 30 specialized AI agents across fraud, AML compliance, credit, onboarding, sanctions, disputes, analytics, workflow, and explainability on a unified data layer: eliminating manual casework, accelerating alert resolution, and connecting risk signals across the entire customer journey

PALO ALTO, Calif., June 3, 2026 /PRNewswire/ -- Oscilar today launched Agent Hub, a coordinated suite of more than 30 AI agents purpose-built for risk operations across fraud, AML compliance, credit, onboarding, sanctions, disputes, analytics, workflow, and explainability. Because Oscilar's unified platform spans the entire customer journey, every agent operates from a richer, more complete data set than point solutions can offer — and agents share signals and context with each other in real time. With every human review, the agents learn, making the entire system sharper over time.

The launch comes as global financial crime reaches $4.4 trillion annually, growing at a 19.2% compound rate, roughly six times faster than the global economy.

Financial institutions are swiftly embracing the unified risk decisioning approach. Three years after emerging from stealth, Oscilar serves more than 100 firms, including SoFi, MoneyGram, Uphold, and Nuvei, processing tens of billions of risk decisions each year.

Oscilar was built on the conviction that you cannot fix fragmented risk infrastructure by adding yet another no-code UI to it. The architecture itself is the problem. That's the bet co-founder and CEO Neha Narkhede made when she started the company. A technical founder, Narkhede co-created Apache Kafka at LinkedIn and co-founded Confluent which scaled to $1 billion in ARR and was acquired by IBM for $11 billion. Her co-founder and CTO, Sachin Kulkarni, spent 11 years at Meta, where he led the team behind Facebook's core infrastructure and the Messenger backend, serving over 1 billion requests per second. The two have built Oscilar without taking outside investment.

The problem: Fragmentation plus engineering dependency drives inefficiency
A fraud flag raised in one system doesn't inform a credit decision in another. AML analysts re-investigate the context that fraud teams have already established, adding hours of duplicated work to alerts that are overwhelmingly false positives. Responding to a new fraud vector means filing an engineering ticket and waiting weeks.

Before building Oscilar, Narkhede spent months speaking to nearly a hundred risk leaders. They told her the same thing: fragmentation plus engineering dependency was the single biggest barrier to fast, accurate risk decisioning. Most "agentic AI" in risk today does nothing to solve this. Single-task agents bolted onto legacy systems inherit the same incomplete, siloed context.

"Fraudsters don't operate in silos. They probe onboarding, test payments, and exploit credit systems as a coordinated attack," said Neha Narkhede, co-founder and CEO of Oscilar. "There's no reason a financial institution's defense should be fragmented when the attack is coordinated. Oscilar Agent Hub orchestrates agents that share intelligence, so risk and compliance teams respond to new threats in minutes, probe their own data without a data team, and build new strategies without writing a single rule or filing a single engineering ticket."

How Agent Hub works
Oscilar's Agent Hub gives compliance, fraud, and credit risk teams a suite of specialized AI agents designed to help them build and deploy new policies 5x faster without engineering support, review alerts at scale 3x faster, reduce false positives by 45%, probe their data using natural language, and proactively recommend changes to risk policies to improve KPIs.

This is powered by more than 30 specialized agents, each operating from the same unified customer risk profile, a continuously updated record aggregating identity signals, transaction history, device and network data, behavioral patterns and prior case outcomes.

Agents in the hub include:

• Workflow Agent: Builds end-to-end risk workflows from natural language descriptions, eliminating engineering dependency for strategy changes.
• Analytics Agent: Lets risk teams query data in natural language and generate dashboards, trend analyses, and recommendations on the fly, without the need for SQL expertise, eliminating dependence on data teams.
• Rule Recommendation Agent: Continuously analyzes rule performance, suggests threshold adjustments, identifies redundant rules, and surfaces new patterns to catch, all backed by data-driven impact analysis.
• AML L1 Review Agent: Reviews first-line AML alerts, surfaces relevant customer history and recommends dispositions, reducing analyst time on the routine reviews that make up 95% of alerts, helping teams to move 5x faster.
• Sanctions Agent: Screens transactions and entities against sanctions lists, surfaces match context and recommends dispositions for analyst review
• Fraud Disputes Agent: Automates insight gathering, reasoning, and recommended action for fraud alerts and disputes, reducing manual review time and increasing accuracy.
• Credit Explainability Agent: Generates human-readable, regulator-ready rationale for every decision and policy change across credit operations.

These agents do more than accelerate routine work. They give every risk operator the ability to build new policies, run advanced analysis, and resolve investigations — work that previously required cross-functional coordination across engineering, data, and compliance. Agents surface recommendations; analysts confirm, adjust, or override them. Every interaction makes the agents sharper. Policy changes that once required a multi-week engineering cycle now take minutes.

Outcomes from Oscilar in production

• SoFi: 50% faster deployment of new credit risk strategies using the Workflow Generator Agent, reducing time from weeks to days, plus a 30%+ improvement in decisioning speed.
• Nuvei: 50% reduction in manual underwriting time with no missed SLAs and a 15% increase in auto-adjudication, all in just the first month.
• Clara: Review time reduced from hours to minutes, with throughput expanding 4x without increasing headcount.

"Workflow Agents support SoFi's ability to make swift and accurate risk decisions and serve our members' needs. We can easily launch and iterate new policies and adapt with unprecedented speed." said Adam Colclasure, Senior Director of Risk Data & Decisioning at SoFi. "We've valued Oscilar's combination of speed, flexibility, and close collaboration as we grow our lending business."

Learn more about Oscilar's Agent Hub at oscilar.com/ai/agents.

About Oscilar

Oscilar is the Agentic Risk Platform for financial institutions, giving risk and compliance teams direct control over fraud detection, AML compliance, credit underwriting, and onboarding without engineering dependency. Agent Hub, Oscilar's suite of more than 30 coordinated AI agents, operates from a unified view of every customer and transaction across the full risk lifecycle, enabling teams to build and deploy policies in plain language, resolve alerts faster, and share intelligence across every risk domain. More than 100 financial institutions trust Oscilar to process tens of billions of risk decisions per year, each in under 100ms. Oscilar is headquartered in Palo Alto, California.

Learn more at oscilar.com.

SOURCE Oscilar