Team Oscilar had quite the week in Las Vegas.

From the main stage to hallway conversations revealing what we can expect from the next decade of finance, Money20/20 USA delivered exactly what this industry needs right now: honest debate about AI, identity, trust, and what it actually takes to build systems that work at machine speed.

I had the privilege of joining two panels: one on stablecoin fraud and another on deepfakes and identity. Both sessions were filled to capacity with standing room only. Alongside Greg Kidd (USBC), Liat Shetret (TRM Labs), Nik Milanović (This Week in Fintech), Naftali Harris (SentiLink), Zhi Zhou (eBay), and Alex Johnson (Fintech Takes), we talked through what it really takes to build trust at machine speed: from stopping irreversible fraud on stablecoin rails to detecting synthetic identities that are growing more sophisticated by the hour.

I also had the opportunity to sit down with Jason Mikula of Fintech Business Weekly for a live podcast recording. In the glass MoneyPot booth with a live audience, we discussed how my journey from creating Apache Kafka to founding Oscilar follows a single throughline: helping organizations make real-time decisions on data that never stops moving.

And to top it all off, we celebrated with the Oscilar team as we won a Banking Award in Identity, AML & KYC. Beyond our platform, that recognition is validation that the industry is ready to move beyond fragmented point solutions toward unified, AI-powered risk decisioning.

Across sessions, hallway conversations, and late-night dinners, five insights crystallized for me:

1. Fraud is evolving faster than defenses

Fraud today operates at machine speed across every channel: mobile apps, stablecoin platforms, P2P networks, card transactions. Attackers use AI to scale synthetic identities, test stolen credentials across thousands of accounts simultaneously, and exploit the milliseconds between authorization and settlement.

Traditional defenses were built for a different era: batch processing that flags suspicious activity hours or days after it happens, rules-based detection that struggles with novel attack patterns, manual review queues that can't keep pace with real-time fraud rings.

In the stablecoin world, this velocity problem becomes existential. Fintechs act as on-ramps between blockchain and traditional banking, inheriting blockchain's speed without reversibility once tokens settle. But the same challenge exists everywhere: mobile wallets settling peer-to-peer payments in seconds, digital banks onboarding customers in minutes, neobanks processing transactions 24/7 across borders.

The infrastructure itself demands a different approach: real-time, adaptive risk systems that reason at the speed of transactions, not the speed of human review cycles.

2. Identity is the new frontier of trust

Stablecoin fraud detection challenges trace back to one missing piece: persistent, verifiable digital identity that travels with users across platforms.

The future of financial trust will rely on privacy-preserving attestations: cryptographic proofs that let institutions verify legitimacy without exposing underlying private data. Think of it as provable reputation: you can demonstrate trustworthiness without revealing your transaction history, social security number, or behavioral patterns.

Trust, in the next era of finance, will be provable, portable, and privacy-preserving. That's not a theoretical concept, it's where the industry will head in the years to come.

3. Real breakthroughs come from unification

Every conversation at Money20/20 circled back to this truth: no single signal can stop modern fraud.

Device fingerprinting catches some threats. Behavioral analytics catches others. Transaction monitoring, identity verification, network analysis, each solves part of the problem. But fraud networks exploit the gaps between these systems.

The companies winning today are unifying identity, device, behavioral, transaction, and network intelligence into one connected intelligence layer. When that intelligence operates in real time, you get machine-speed trust: the ability to decide in milliseconds what's safe, what's risky, and what requires human judgment.

At Oscilar, this has been our thesis from Day 1: risk decisioning requires a unified platform, not a collection of disconnected tools duct-taped together with manual workflows.

4. Risk isn't a fintech problem, it's a data and AI problem

Every fraud alert. Every credit approval. Every AML check. At its core, each is a real-time data decision under uncertainty.

What I saw across financial institutions was fragmentation: siloed tools solving small pieces of the problem, creating blind spots at every handoff. Teams spend more time wrestling with infrastructure than actually fighting fraud.

Oscilar was born from a simple premise: if Kafka/Confluent built the highways for data movement, risk decisioning is the intelligent traffic control; reasoning in milliseconds over what's trusted, what's risky, and what requires escalation.

That's why we built a real-time, AI-native platform from the ground up, designed specifically for the velocity and complexity of modern financial risk.

5. AI that works is built on explainability, governance, and speed

There's no shortage of AI hype, but in risk decisioning, the value is measurable and immediate:

• Pattern recognition across billions of signals to detect fraud networks that humans miss: the subtle correlations across devices, behaviors, and transaction patterns that reveal organized attacks.

• AI co-pilots that let analysts write, interpret, and explain rules in natural language, turning a conversation (e.g. "flag transactions over $5,000 from new accounts in the past 48 hours with mismatched billing addresses") into executable logic without coding.

• Operational efficiency that reduces false positives by up to 90%, so humans can focus on the 10% of cases that truly require human judgment.

The future of trust won’t be built by replacing people with AI, it will be built by empowering them with AI co-pilots that are fast, explainable, and governed by human oversight.

A week of connection and collaborative thought

Beyond the main stage, we hosted events that brought together the people actually building the future of financial infrastructure:

• Golf Classic with iDENTIFY & FS Vector at Wildhorse Golf Club

• Banking Executive Dinner at estiatorio Milos

• Fintech Executive Lunch with iDENTIFY, FS Vector, and the American Fintech Council

• Roadshow to Stablecon: Stablecoin Day

• Oscilar x TWIF Fraud & Risk Happy Hour at HaSalon

• Data & Darts Happy Hour at Flight Club Las Vegas

A hearty thanks to all who co-hosted and showed up strong, particularly SoFi, Uphold, CCBank, COREBANK, FS Vector, iDENTIFY, American Fintech Council, Kaufman Rossin, and SentiLink. The conversations that happen over dinner, between sessions, at the golf course, are where some of the best work in building trust infrastructure happens.

As I reflect on the week, one thing stands out in my mind: that the next generation of financial trust will be co-created. 

It will take banks, fintechs, regulators, and infrastructure builders working side by side — with AI as the connective tissue — to build systems where trust is instant, explainable, and scalable across billions of transactions.

That's the work ahead. And after this week, I'm more convinced than ever that we're building it with the right people.

Here's to that future.