This ex-Pleo exec plans on taking on US rivals with his ‘AI CFO’ startup

The Unfinished Business of FinTech

For a long time, the promise of fintech was simple: make spending money easier. We got slick metal cards, mobile apps that didn't crash, and instant expense reporting. But for the founders and CFOs sitting in the back office, the core problem never actually changed. We just got faster at making a mess of our books.

Jonathan Sanders, who spent years scaling the Danish unicorn Pleo, saw this firsthand. You can give every employee a corporate card, but you're still left with a fragmented pile of data that doesn't talk to itself. Most startups are running their multimillion-dollar businesses on a fragile web of spreadsheets and basic accounting software that requires constant human babysitting. Sanders recently launched Light to try and kill that manual labor once and for all.

The goal isn't just another dashboard. It's a fundamental shift from reactive accounting to what he calls an AI CFO. If you are building in this space, you need to pay attention to why this is happening now, and why the old guard is finally vulnerable.

Why General Ledger Software is Broken

If you have ever tried to integrate a modern tech stack with a legacy ERP (Enterprise Resource Planning) system, you know the pain. These systems were built decades ago as rigid databases. They were never designed to handle the velocity of a global startup, multiple currencies, or the sheer volume of SaaS subscriptions we all carry now.

Sanders is betting that the AI era allows for a ledger-first approach. Most AI tools in finance today are just thin wrappers. They sit on top of your existing messy data and try to summarize it. Light is trying to be the source of truth itself. For builders, the takeaway is clear: the value isn't in the AI model you use; it's in the architecture of the data the model is allowed to touch.

The current workflow for a finance team is essentially data entry disguised as high-level analysis. You spend 80% of your time reconciling transactions and 20% actually thinking about how to grow the company. Light wants to flip that ratio. By using LLMs to categorize, flag, and reconcile in real-time, the system moves from being a historical record to a predictive tool.

The Multi-Entity Nightmare

One of the biggest hurdles for growing startups is international expansion. The moment you open a subsidiary in a different country, your complexity doesn't double—it quadruples. You have different tax laws, different currencies, and different banking rails. Legacy software usually handles this by forcing you to open separate accounts and manually consolidate them at the end of the month.

This is where the "AI CFO" concept actually starts to make sense for founders. Sanders is positioning Light to handle multi-entity management natively. Instead of waiting three weeks for a month-end close to realize you overspent in your UK office, the system sees the global picture instantly. It's not about being "smart"; it's about being fast and connected.

For those of us in the crypto and AI space, we see this pattern everywhere. The bridge between the old world (manual reconciliation) and the new world (automated intelligence) is always high-quality, structured data. If you can't trust the ledger, the AI is just a glorified hallucination engine.

The Competitive Landscape: US vs. Europe

Sanders isn't walking into an empty room. In the US, companies like Brex and Mercury are moving aggressively into the software layer. They started as cards and banks, and now they want to be the brain of the company. However, Europe presents a unique challenge—and a unique moat. The regulatory fragmentation here is a nightmare, which is exactly why a European-born founder might have an edge in building a truly global system.

We are seeing a trend where the "CFO stack" is becoming the most contested territory in SaaS. Why? Because that's where the data gravity is. If you control a company's ledger, you control their decision-making process. Every other tool—from HR to marketing—eventually has to report back to the money.

A Founder's Perspective on Skepticism

I’ve seen dozens of "AI-powered" finance tools cross my desk. Most of them are vaporware. They promise to automate your taxes, but in reality, they just send you an email when they get confused. The skepticism here is healthy. As a founder, you cannot afford to have an AI hallucinate your burn rate or miscalculate your runway.

Sanders seems to understand that the human element doesn't go away; it just changes roles. The AI handles the drudgery, but the human sets the strategy. The real test for Light will be whether it can maintain 100% accuracy while scaling. In finance, 99% accuracy is a failure.

The value of AI in finance isn't in its ability to write poetry; it's in its ability to never get tired of looking at 50,000 lines of transaction data.

What This Means for Builders

If you are building in AI or FinTech right now, the "Light" approach offers a few lessons:

• Don't be a wrapper: If you are just putting a UI on top of someone else's broken data, you will be disrupted. You have to own the underlying record.
• Solve for complexity, not just speed: Speed is a feature, but managing multi-entity complexity is a moat.
• Vertical AI is the future: Horizontal AI (like ChatGPT) is great for brainstorming, but vertical AI that understands the specific rules of accounting is where the enterprise value sits.

We are moving away from the era of "software as a service" and into "software as an employee." A tool that can autonomously manage a ledger isn't just a tool; it's a team member. Sanders and his team at Light are essentially trying to build a digital employee that never sleeps and never misses a decimal point.

The Bottom Line

The race for the AI CFO is on. While the US rivals have more capital, the complexity of global markets requires a more nuanced, data-first approach. Jonathan Sanders is betting that his experience at Pleo gave him the roadmap to avoid the traps of legacy finance. Whether Light becomes the new standard or just another tool in the stack depends on how well they can prove that their AI is more reliable than a human junior accountant. In the world of finance, trust is the only currency that matters.

Read the original at Sifted →