Fravity AI Joins Socure's RiskOS Agentic Stack

Fravity AI built agentic software for the work that begins after a fraud system raises its hand. Its platform gathered evidence, ran screening, reconstructed cases, and prepared investigator-ready documentation for fraud, risk, AML, and compliance teams.

Founded in Austin in 2024 by Kedar Samant and Rushik Upadhyay, Fravity moved quickly from a small specialist startup into the operating layer of a much larger risk platform. On August 27, 2026, Socure acquired Fravity for undisclosed terms and announced that its technology would become RiskOS_Agents inside Socure's RiskOS platform.

The broader implication reaches beyond one acquisition. Identity and fraud platforms have spent years improving the decision that creates an alert, while investigators still absorb the slower work required to explain, document, and resolve it. Fravity made that second workload its product, and Socure is now connecting it to a network processing roughly 10B decisions a year.

About Fravity AI

Fravity was a privately held company with a LinkedIn-listed team size of 2-10 before the acquisition. Its mission centered on giving operations teams agentic tools that could handle repetitive investigative work while preserving human control, explainable reasoning, and audit trails.

The platform combined 2 product surfaces. Agentic Copilot brought prebuilt agents into an investigator's existing workflow, while Agentic Studio let teams assemble and customize multi-agent processes. Fravity described more than 70 agents spanning sanctions and politically exposed person screening, adverse media, document intelligence, business due diligence, transaction-monitoring triage, and case documentation.

That design matters because regulated investigations rarely live inside one clean database. A single alert can require internal transaction history, customer information, external watchlists, corporate records, documents, prior cases, policy logic, and an explanation that another analyst or auditor can reconstruct later. Fravity's product tried to turn that scattered work into a repeatable workflow without pretending human judgment had become optional.

The Problem Fravity AI Was Solving

Fraud detection software is good at producing risk scores and alerts. Operations teams still have to determine what happened, whether the signal is material, what evidence supports a disposition, and what action should follow. That gap between detection and resolution is expensive because it consumes experienced people one case at a time.

Fravity positioned agents as a way to compress the mechanical parts of that work. The software could retrieve information, apply screening steps, enrich a case, draft a narrative, and preserve a record of how the result was assembled. The operating goal was capacity: help investigators handle more volume and complexity without making headcount rise in a straight line with alert volume.

The company reported that existing deployments reduced cost per case by up to 80%, accelerated case resolution by up to 5x, and lowered false positives by as much as 70%. Those figures are company-reported outcomes, not independent benchmarks. A more concrete commercial signal appeared in the acquisition announcement, which said several enterprises were already running Fravity and Socure together in production.

Leadership Built Around Fraud Operations

Kedar Samant brought a long fraud-platform history to Fravity. The company's official biography said Samant built Google's global fraud-management platform across Ads, Commerce, and Social products, then co-founded Simility. PayPal agreed to acquire Simility for $120M in 2018, and Samant later served as Sr. Director of Fraud & Risk at PayPal.

Rushik Upadhyay served as Fravity's co-founder and CTO. Fravity's biography and a 2025 industry presentation described nearly 2 decades of work at PayPal, including architecture for its AML and financial-crime platform. Together, the founders approached agentic AI from inside the regulated queue rather than from a generic automation thesis.

That domain history shows up in the product's emphasis on controls. Fraud and compliance teams do not need an agent that merely sounds confident. They need traceable inputs, repeatable workflows, escalation paths, explainable outputs, and a record that can survive review by another operator, an auditor, or a regulator.

From Fravity to RiskOS_Agents

Socure says Fravity's capabilities will be delivered through RiskOS as RiskOS_Agents. The acquired technology adds an agent-building and ontology layer to Socure's identity, fraud, authentication, and compliance platform, connecting investigative workflows to proprietary data, purpose-built models, and downstream case outcomes.

The scale changes the engineering problem. Socure serves more than 3,000 customers across more than 190 countries and says its network processes roughly 10B decisions a year. Fravity's independent product thesis now has to work across a much broader range of customers, policies, data environments, and regulated use cases.

The two companies already shared customers, which gives the integration a practical starting point. The next proof will come from whether RiskOS_Agents can preserve Fravity's investigation depth and controls while gaining the distribution, data feedback, and platform reach of Socure.

Why Hiring Momentum Now Sits With Socure

Fravity's website now redirects to Socure, and a standalone Fravity careers page is no longer verifiable. That makes it inaccurate to present Fravity as an independent company recruiting across its old organization. Current opportunities belong to Socure and should be verified through the Socure careers page.

Socure is hiring for work adjacent to the acquired thesis. A current AI Engineer role includes production agent design, orchestration, human-in-the-loop controls, evaluation, observability, and agentic foundations. The hiring signal is less about adding bodies to a small startup and more about industrializing agent workflows inside a large identity and risk platform.

For builders, that is a harder and more useful assignment. Fraud and compliance agents need to reason across messy data while remaining measurable, reviewable, and secure. The people joining this work will decide whether agentic systems become credible operational infrastructure or another layer that investigators must monitor by hand.

What Fravity AI Signals for Fraud and Compliance

Fravity's short independent life captured a change in the fraud stack. Risk platforms are moving from verification and scoring toward an operating loop that can detect, decide, investigate, document, and learn from the resolution. The acquisition puts that loop inside one platform with a large customer network and a growing library of outcomes.

The market pressure is straightforward. AI can increase both the scale of fraud and the capacity of defensive teams, but the defensive advantage depends on controls, data quality, and the ability to explain what an automated system did. Fravity's contribution was to treat investigation work as a product surface with its own architecture, not as the paperwork left over after a model returns a score.

The Fravity brand may recede as RiskOS_Agents takes its place. The operating burden that created the company remains visible in every alert queue, and Socure now has to prove that the acquired system can carry that burden across institutions without losing the traceability that made it useful.