Socure Buys Fravity: Agentic AI Closes the Fraud Loop
On August 27, 2026, Socure announced a $156 million strategic growth investment at a $5.2 billion valuation and, in the same breath, the acquisition of Fravity, an agentic AI platform for fraud and compliance operations. The combined move closes a gap that has quietly defined the limits of every fraud-detection system: the model can flag the problem, but a human still has to investigate it.
That bottleneck is now the target.
The Problem Every Enterprise Fraud Team Knows
Enterprise fraud teams are caught in a structural bind. Better AI detection means more alerts. More alerts mean more cases requiring human review. But analyst headcount does not scale linearly with fraud volume, and AI-powered attacks are accelerating faster than any hiring plan can match.
Intelligence platform Liminal puts hard numbers on the problem: U.S. organizations spend $100 billion a year staffing internal and outsourced fraud, compliance, and risk management teams. Fifty-three percent of banks spend at least an hour reviewing each alert. Thirty-seven percent manually review more than 40% of their alerts. And across the industry, AI-driven fraud attacks rose 8,000% in the past year, creating alert volumes growing faster than compliance teams can staff against.
The math simply does not work for the manual-review model. Socure’s acquisition of Fravity is a direct answer to that arithmetic.
What Fravity Actually Built
Fravity, founded in 2024 and based in Austin, built an agentic operations platform that deploys AI agents against the investigative queues that compliance and fraud analysts currently handle by hand. When a transaction-monitoring system produces a suspicious-activity alert, a human analyst typically has to pull documentation from multiple systems, cross-reference watchlists, review transaction history, conduct adverse-media searches, and draft a case file, all before making a disposition decision. The process consumes hours per case and scales linearly with alert volume.
Fravity’s agents run those steps end-to-end. According to Biometric Update, the platform covers KYC investigations, business due diligence, transaction-monitoring alert assessment, sanctions and PEP screening, and adverse-media research. The Agentic Studio module gives enterprise teams access to more than 70 pre-built agents, with the option to build custom agents around internal risk policies and regulatory requirements. The Agentic Copilot module executes workflows and produces explainable, case-ready outputs for analyst review.
Critically, Fravity does not remove the human from the final disposition decision. It removes the human from the evidence-gathering and documentation work that precedes that decision. Analysts concentrate on judgment; agents handle the paperwork.
Across shared enterprise customers running both Socure and Fravity in production, the results have been: 80% reduction in cost per case, five-times faster case resolution, and up to 70% fewer false positives. Those figures come from Socure’s official BusinessWire announcement and represent live deployments, not benchmarks.
Why the Acquisition Architecture Matters
Socure and Fravity did not come together as strangers. The two companies already share multiple enterprise customers running both platforms in production, and the founding teams have worked together across multiple companies for more than a decade. That pre-existing customer overlap means the integration does not begin in the lab; it begins with live production data from day one.
Fravity’s capabilities will be delivered through Socure’s RiskOS platform as RiskOS_Agents. The structural advantage of that integration is data: RiskOS_Agents will learn from approximately 10 billion decisions per year across Socure’s network, grounding every agent action in real outcomes from 3,000-plus enterprise deployments. Standalone agentic vendors have no path to that feedback loop.
As Socure CEO Johnny Ayers put it: “AI broke the economics of fraud and compliance operations, and no institution is going to hire their way out of it. The next decade belongs to the organization who owns the full loop: the data, the models, the decision layer, and now agents that act on all three components.”
The Investment and What It Signals
Summit Partners led the $156 million raise, with Goldman Sachs Alternatives, Wells Fargo, and Docusign also participating. The round mixes primary capital with an employee secondary tender offer. The $5.2 billion valuation represents a 16% step-up from Socure’s last mark, set in its November 2021 Series E, and comes after a period of accelerating fundamentals.
Socure closed Q2 2026 with $364 million in annual recurring revenue, 63% above the prior-year figure, 133% net dollar retention, and 0.01% logo churn. The company added 95 customers in the quarter, including Circle, Cox Automotive, MoneyLion, and Login.gov, and now serves more than 3,000 customers across more than 190 countries, including 18 of the 20 largest U.S. banks.
Andy Collins, managing director at Summit Partners, framed the investment thesis clearly: “Identity has become a primary control point for trust in an AI-driven economy, and we believe the platforms that can verify identity accurately at global scale will define the next decade of risk infrastructure.”
The timing matters. Enterprise AI is accelerating the sophistication and volume of financial crime simultaneously. Organizations that can build the full loop (detection, investigation, and resolution, all within one governed platform) have a structural advantage over those stitching together point solutions.
What Enterprise Leaders Should Take From This
The Socure-Fravity deal is a data point in a pattern that is now visible across the enterprise AI market. As our analysis of the Salesforce 2026 Agentic Enterprise Index showed, the enterprises deploying agentic AI at scale are increasingly measuring success not by whether agents can perform a task in isolation, but by whether they can be integrated deeply enough into systems of record to act without hand-holding.
Fraud investigation is one of the clearest examples. The detection layer, (identity graphs, transaction scoring, device intelligence) already runs without humans. The resolution layer, (case building, evidence assembly, documentation) still runs on analyst hours. Fravity’s Agentic Studio closes that gap by wiring agents directly into the workflow between alert and disposition.
The broader implication for enterprise AI strategy: the next phase of agentic ROI is not building more autonomous capabilities from scratch. It is acquiring or integrating proven agentic platforms that already carry production-validated outcomes. Socure paid an undisclosed acquisition price for Fravity but received a platform with live 80% cost reductions in customer deployments. That is a very different calculus from buying potential.
For security and compliance leaders, the adjacent governance question is also worth watching. As Zenity’s $125M round in August 2026 highlighted, every new agentic layer inside an enterprise introduces new questions about agent identity, permissions, and audit trails. RiskOS_Agents operates inside a heavily regulated context where explainability and auditability are not optional. How Socure handles the governance layer for its agents will be a case study for every team deploying agentic AI in financial services.
Fraud Investigation by the Numbers
| Metric | Current State | With Fravity Agents |
|---|---|---|
| Cost per investigation case | Baseline | 80% lower |
| Case resolution time | Baseline | Up to 5x faster |
| False positive rate | Baseline | Up to 70% lower |
| Alert volume trend | Rising 8,000% YoY (AI-driven fraud) | Agents absorb the surge |
| Bank analyst time per alert | 53% spend 1+ hour per alert | Agents handle evidence gathering |
| Manual review share | 37% of banks review 40%+ of alerts manually | Reduced to judgment-only exceptions |
Sources: Socure/BusinessWire, Biometric Update, Liminal via Finovate
The Road Ahead
Socure says initial RiskOS_Agents modules target watchlist screening, monitoring, and KYB checks, with additional offerings planned across financial services, workforce and payroll, public sector, insurance, gaming, and crypto.
The open beta for RiskOS_Agents has no published timeline beyond the initial enterprise rollout. For institutions already running Socure’s RiskOS platform, the integration will be native rather than a separate procurement, which reduces deployment friction compared with adopting a standalone vendor.
For the broader enterprise AI market, the announcement is another signal that the agentic layer is moving from experimental to infrastructure-grade. Organizations spending analyst hours on alert investigation today are operating a cost structure that AI-enabled fraud volumes will make unsustainable. The question is not whether to automate investigation workflows. The question is which platform will own that layer inside your stack.
If you are building an enterprise AI strategy that includes agentic automation for finance, compliance, or operations, Enera can help you design the right architecture.