Financial compliance is everything AI promised to automate. It's repetitive, rules-based, and burns thousands of analyst hours copying data between systems, chasing documents through email threads, and drafting reports no one wants to write.

Chrisjan Wüst lived this problem as head of compliance at RelyComply, a South African compliance startup. When he met Alexandre Berkovic, who comes from a family of artists, they initially built a startup in the creative space. But working together, they realised the bigger opportunity was automating Chrisjan's old job.

The gap was clear. RegTech companies added AI features to dashboards, but analysts still did all the actual work.

This is exactly the type of opportunity Cherry looks for: an AI-native company tackling a massive, fragmented industry. We're leading Sphinx's $7.1M seed round with participation from Y Combinator and others.

The Compliance Reality

Financial institutions are legally required to verify who their customers are and ensure they're not facilitating money laundering or terrorist financing. This is Know Your Customer (KYC), Know Your Business (KYB), and Anti-Money Laundering (AML) compliance.

Banks must check identities against sanctions lists, monitor transactions for suspicious patterns, investigate alerts, gather supporting documentation, and file reports with regulators when something looks wrong.

The problem is this doesn't live inside any single platform. It lives across email, PDFs, third-party web portals, legacy banking systems, and government databases that were never designed to talk to each other.

Analysts spend their days as human glue between systems, copying information from one place to another, tracking down documents, and manually assembling regulator-ready reports.

This is why adding AI to one vendor's dashboard doesn't work. The AI can't reach the email. It can't log into the third-party portal or pull a PDF from a customer conversation or check the government registry. Current solutions sit inside a walled garden while the actual work happens everywhere else.

Because of this, financial institutions spend over $200 billion annually on compliance teams and outsourced review doing work software was never built to handle. According to Interpol, despite this spending, institutions still detect only around 2% of global financial crime - because the volume is impossible to handle manually and the tools don't actually do the work.

How Sphinx Works

Sphinx agents work directly inside the environments where human analysts currently operate.

They log into existing case management systems, third-party portals, email, and internal dashboards. They review alerts, conduct AML and KYB checks, gather supporting research from multiple sources, draft requests for information, track responses, reconcile everything, and file regulator-ready reports. The system captures the full reasoning process in audit trails that regulators can inspect.

This approach means institutions can go live in days without replacing existing systems. The agents adapt to whatever tools are already in place, understand what they're looking at, make decisions based on regulatory requirements, and document their work in ways that satisfy auditors.

Deployment is staged to build that trust. During the first two to three months, all cases run with human-in-the-loop review. An internal quality assurance team staffed with former compliance analysts reviews 50-70% of cases during this phase, providing feedback loops for the models and the level of accountability regulators require.

As the system accumulates training data from approval and rejection cycles, it gradually increases the proportion of cases handled autonomously. Current benchmarks show 85-90% accuracy depending on case type, with the QA process covering the remaining gap.

The results have been dramatic.

Equals Money saw a 94% reduction in false positives while catching more true positives. Several institutions now operate internationally without expanding compliance headcount, cutting operational costs by up to 4x. In production, Sphinx agents have handled millions of alerts and hundreds of thousands of cases, clearing months-long backlogs in days. Customers report faster onboarding, materially lower exposure to compliance errors, and the ability to scale operations without hiring additional analysts.

Why We Backed Them

We met Alex and Chrisjan at Cherry's Engine Room dinner in Paris in spring 2023 when they were working on their creative AI startup.

But with Chrisjan’s background, compliance was always in the back of their mind. When they didn't get into Y Combinator initially, they regrouped and applied again under the compliance thesis. Through that process, Alex went deep on the problem and found genuine passion in solving it.

Two things convinced us.

1. First, watching Alex develop real conviction around compliance as he understood how it actually worked and saw what was possible that no one was building.
2. Second, the product makes end-to-end decisions, not just recommendations. It runs fully autonomously without requiring humans to review every case unless there are edge cases. That's the difference between a tool and a replacement.

As our partner Filip Dames recently wrote, software is being revalued based on whether it executes outcomes or coordinates work. Sphinx executes.

When enterprise contracts started coming through across multiple geographies, the inflection point was obvious. By the time we invested, Sphinx already had customers across more than a dozen countries. Banks, public companies, and fast-growing fintechs were using Sphinx to eliminate manual review work and scale operations globally. The company needed capital to fulfill demand.

Meanwhile, the cost of non-compliance is surging. In the first half of 2025 alone, regulators issued 139 penalties totaling $1.23 billion, a 417% increase versus the previous year. AI-driven fraud grows more sophisticated. Institutions need compliance operations that can keep pace, and hiring more analysts isn't sustainable.

The team, composed of second time founders, ex-operators and former compliance officers, is positioned to fulfill AI's promise to automate boring work, finally delivered for one of the most obvious use cases. We're backing Sphinx to become every financial institution's last compliance hire.