Oscilar

We went to CAFC's Agentic AI in Financial Crime conference in New York, and no one was asking whether to use agents. They were asking how to prove the agents work, who's accountable when they don't, and whether to build or buy the next one. That last one in particular caught a lot of attention. Swipe for the top 10 trends. →

A lot of the looming AI safety debate has centered on doomsday scenarios and a distant 2030. Neha Narkhede's point to Jon Swartz was closer to the decisions companies are making today: “We should be willing to slow advances whose risks we cannot adequately assess or control, while continuing to deploy useful applications whose performance and operating limits we can demonstrate. Each deployment has to meet that standard, including tools intended to protect people.” In risk and compliance, that standard is already familiar: prove how a model performs, where it fails, and who owns the decision before you ship it.

Earlier this month, Sachin Kulkarni and Jonathan Klingler joined Shawn Budde of Ensemblex, who was putting machine learning into production underwriting well before today's AI wave, and a table of credit risk leaders for the second conversation in this series. The discussion focused on a few hard questions: - Where agents are already doing meaningful work in underwriting, and where they still break down - What model risk teams and regulators actually require, versus what vendor decks imply - Whether adding more data consistently leads to better decisions - Which gaps in auditability, controls, and infrastructure the industry has yet to fully account for As agentic systems move closer to real credit decisions, the pressure is moving towards what institutions can explain, govern, audit, and defend in production. Thank you to Ensemblex for hosting and to everyone who joined the table and brought the candor.

Most fintechs on the charter path already have an AML program. The question a bank regulator asks is a different one: is it the program a chartered bank needs, and can you show the work? Under a sponsor bank, a good share of the compliance program is technically someone else's concern. A charter moves all of it onto your books at once, and the standard goes up on the way over. Tomorrow at 10 AM PT, Evey Guo of FS Vector, Syed Razaand of FTI Consulting, and Phil Goldfeder of the American Fintech Council join Saurabh Bajaj for an hour on whether and which charter to pursue, what regulators expect to see before you file, where a sponsor-era fraud and AML setup may fall short, and what the de novo period looks like once you're open. Register below.

"Fraud doesn't wait for this debate to end." Our co-founder and CTO Sachin Kulkarni spoke with InfoMoney about cloned voices, forged documents, and synthetic identities that are cheap to buy and easy to run at volume. You know, the AI risk that's already here, not the one in the headlines. His position on using AI to fight it: yes, but with rigorous testing, narrow permissions, continuous monitoring, and a human who stays accountable for the call. Read the full article at the link below. Thanks to Maria Luiza Dourado for the conversation.

We're not saying Booth 221 is lucky, but you'd be hard pressed to find better odds in Vegas. Don’t miss your chance for some face time with Seth Sattler, Seshu Bommineni, CAMS, Matt Pless, Saurabh Bajaj, and Timothy Lam at ACAMS: The Assembly. They won’t be there to lecture you 101-style. But they will answer your real questions. So bring them. And if you're free the evening of Tuesday, Sept. 29, we're hosting an exclusive dinner at Carbone, 5–7pm. Learn more and request your seat below.

I spoke with Maria Luiza Dourado at Brazilian outlet InfoMoney about AI's dual role in perpetuating and defeating fraud. Much of today's AI conversation focuses on frontier models and whether labs should slow down. Those concerns are serious, and we share many of them. But financial institutions are already dealing with AI-powered threats. They need to evaluate new technology against their own data, policies, and controls while attacks are underway. Some of the biggest exposure is hiding in plain sight. We estimate that 5-7% of what banks record as defaults or credit losses is actually undetected fraud, misclassified as a credit problem. Traditional fraud signals are becoming less reliable in isolation. Device fingerprints, behavioral scores, and velocity checks still matter, but increasingly, it's the patterns across thousands of sessions and the entire customer journey that reveal what's actually happening. At Oscilar, we see AI as essential to making that connected view possible at scale. Investigators bring the context and judgment to decide what those patterns mean, and retain accountability for the decision. ------------------------------ Conversei com a Maria Luiza Dourado, do InfoMoney, sobre o papel duplo da IA: ao mesmo tempo em que alimenta a fraude, também é essencial para combatê-la. Boa parte do debate atual sobre IA está concentrada nos modelos de fronteira e em se os laboratórios de ponta deveriam desacelerar. Essas preocupações são sérias, e compartilhamos várias delas. Mas as instituições financeiras já estão lidando com ameaças impulsionadas por IA. Elas precisam avaliar novas tecnologias frente aos próprios dados, políticas e controles, enquanto os ataques estão em curso. Parte da maior exposição está escondida à vista de todos. Estimamos que 5% a 7% do que os bancos registram como inadimplência ou perda de crédito é, na verdade, fraude não detectada, classificada erroneamente como problema de crédito. Os sinais tradicionais de fraude estão se tornando menos confiáveis isoladamente. Fingerprint de dispositivo, scores comportamentais e verificações de velocidade continuam relevantes, mas cada vez mais são os padrões observados em milhares de sessões e toda a jornada do cliente que revelam o que está realmente acontecendo. Na Oscilar, vemos a IA como essencial para tornar essa visão conectada viável em escala. Os revisores e tomadores de decisão trazem o contexto e o julgamento para decidir o que esses padrões significam, e mantêm a responsabilidade pela decisão final.

Under a sponsor bank, the answer to "is this enough?" was partly someone else's to give. Is the AML program you run under your MTLs enough for a bank regulator? How much of your management team has to be hired before you file? Can you dual-hat people across the bank and the parent company? Under your own charter, you own every one of those answers, and you defend them at every exam. @Evey Guo of FS Vector and Phil Goldfeder of the American Fintech Council join our CPO Saurabh Bajaj this Thursday to get into it: choosing the right charter type, the governance and staffing regulators expect before and after filing, building fraud and AML for a chartered model, and what ongoing examination looks like once you're open. 📆 Thursday, September 24 ⏰ 10 AM PT / 1 PM ET Save your seat at the link in the comments.

If you work in fraud or financial crime in New York, you know the drill: the controls catch what they were built to catch, and everything else shows up in your queue. Come talk about everything else over breakfast. Oscilar, SentiLink, and FS Vector are co-hosting a small, off-the-record panel on Thursday, Sept. 24, 9:00–11:00AM ET at FS Vector's office in the Flatiron District. Coffee, breakfast bites, and two hours of the conversation people usually only have off stage. What's on the table: - Where identity fraud is still getting through - Why the review queue never actually empties - How much of this work AI can take on, and what a regulator or sponsor bank expects to see behind an automated decision Full speaker lineup coming soon. Seats are capped, so the sooner you raise your hand the better. Link below.

Most AML monitoring systems were built around the shape of a bank: one institution, one jurisdiction, and customer activity that can be followed from one end to the other. dLocal (NASDAQ: DLO) operates in a much more complex environment. 60+ markets across Africa, the Middle East, Asia, and Latin America. More than 1,000 payment methods. Different monitoring requirements, different data access rules, different licensing regimes, and payment flows that cross multiple parties and borders. And the environment keeps changing. Volumes shift and new licenses come online, so the assumptions underneath the compliance program have to change with them. As Christiana Ellina, dLocal's VP of Compliance, put it: "Traditional bank-oriented monitoring systems were never designed for this environment." After years of building infrastructure, raw volume is rarely the hardest problem. The harder problem is staying correct across a growing number of distinct conditions. With Oscilar, compliance teams can change rules and thresholds directly, without waiting on engineering. Agents handle the heaviest parts of investigations while analysts retain judgment. Every change leaves an audit trail an examiner can follow. dLocal is now running Oscilar across AML transaction monitoring, sanctions screening, and case management. Behind "One dLocal," one API and one contract for merchants, is a compliance operation absorbing enormous variation across markets. That is exactly the kind of challenge Oscilar is built to handle.