Behind the Stack : How Defacto is Rebuilding B2B Credit Infrastructure

Welcome back! For the second episode of Behind the Stack, we sat down with Marco Gires, Defacto’s CTO to dive into the tech powering one of Europe’s most ambitious embedded lending platforms. This year they raised a €16M equity round and expanded their SME working capital financing capacity to €1.8 billion, reaching a milestone of lending to 12,000 SMEs with over €1 billion in financing. 

From API-first design to real-time risk management, Defacto has spent the past two years rethinking what B2B credit infrastructure can look like. In this conversation, we explore how their engineering culture shapes product strategy, why compliance is a moat, and the bold bets that helped them scale fast with a lean team.

1. How would you describe Defacto’s technology vision, and how has it evolved over the past 2 years?

When we started Defacto, the mission was bold: become the first truly API-first instant lending solution for SMBs’ working capital. To do that, we had to rebuild everything from onboarding to disbursement, because speed only comes when you own the full value chain.

Fast forward two years: we've not just proven that instant lending works, we've rewritten the economics of B2B credit. And now? Others want to build on top of what we've created. So we’re evolving into something bigger: a modular, scalable lending infrastructure that allows any partner to build and distribute tailor-made financial products, anywhere.

2. What role does your engineering team play in shaping the product strategy?

Tech runs through Defacto’s veins. Every business unit has a single owner who drives it, but always with tech at their side. We’re a lean team of under 30, but over half of us are senior engineers with 7+ years of experience. That gives us massive leverage and lets tech shape not just how we scale, but where we go next.

3. How do you balance long-term innovation with short-term delivery pressure?

I try to embed long-term thinking into every short-term decision. It’s like planting a forest: every quick win should grow within the bigger design.

My golden rule? Once a problem is well-scoped and the design is solid, implementation is just... CRUD. Fast, simple, reliable.

4. What architectural principles guide your platform design?

We live by three core principles:

Domain-Driven Design → Clear boundaries = better scale.

API-First → Everything we do must be programmable.

Critical vs. non-critical path → Lending has only two truly critical moments: onboarding and loan submission.

Everything else, fraud checks, document parsing, scoring, happens asynchronously in the background.

5. How do you support real-time embedded lending at scale?

Real-time lending isn’t just about speed, it’s about trusting a machine to say “yes” to €9M in a single day. So you need governance baked into the system. Can you explain every decision? Can you monitor its behavior live? If something changes, can you adjust the logic in minutes?

If you can answer yes to all three, then you’re ready for real-time lending.

6. Any bold tech bets that paid off (or didn’t)?

We went all in on LLMs across our stack, and it's been game-changing. Despite staying at 27 people, our revenue per employee is €450k. LLMs help us scale ops, risk, customer care… everywhere. It’s like having a superpower baked into every team.

7. How is your team structured for collaboration?

No rigid org charts here. We shift structure based on what matters. For a major launch, everyone on deck. For scaling, split by business unit. For deep technical work, align by expertise (security, infra, etc.).

But if I had to pick one structure we come back to, its business-unit squads: risk, distribution, payment infra, etc. We hire generalists who can move across units. Because moving builds understanding, and understanding drives better decisions.

8. What’s your hiring philosophy?

We never hire for a specific role. Everyone we bring in is fullstack, flexible, and curious, because the startup roadmap changes fast.

Also, we never forget: great candidates are also interviewing us. We work hard to be the place they want to choose. Our process is designed so that we’re their #1 option.

9. How do you handle data governance in embedded lending?

Honestly, this is one of the most complex parts, partly because legal clarity isn’t always there.

So we built our own data access layer, tracking ownership and rights for every record. Every request carries an identity, and that’s enforced at the DB level. It's precise, auditable, and scalable and can be configured to fit the legal requirements.

10. How do you balance compliance and innovation?

First, design the best user journey. Then reverse-engineer regulation into it.

Also, a truth we learned early: compliance is a moat. It makes you more resilient.

We’ve embedded almost all regulatory processes into our platform. Why? Because when you’ve had 4 independent audits in 6 months, manual processes stop scaling.

11. How do you embed fraud and credit risk into the stack?

We’ve centralized all fraud logic into one domain. No new loan gets greenlit if there’s a suspicion, it’s systemic.

Regarding tooling, operators and services can easily flag suspicions, creating a robust, auditable dataset.

And on the credit risk side, we separated the model from the underwriting logic. The model gives data, the logic makes the call.

Depending on complexity, models live in Git, SageMaker, or 3rd-party platforms.

12. How do you design APIs and integrations for embedded finance?

API design is about agnosticism, don’t hardcode today’s use case. Our APIs can support credit cards, BNPL, long-term loans... the works.

For integrations, we often work with integrators. Why? Because our core IP is lending, not connectors. Our edge is in underwriting and automation.

13. What’s hardest about plugging into partner ecosystems?

Three things come to mind. Partner dev time is scarce, so changing anything post-launch is a negotiation. On the marketing front, even with a killer integration, partners might not promote it. And the balance between eligibility vs. value, meaning partners want wide eligibility. Lending often means narrowing access. Finding that balance is tricky.

14. How do you choose external platforms (KYC, aggregators, etc.)?

We take into account a number of factors but the main ones being API quality, support responsiveness, international reach, and cost (for thin-margin lending, €10/month/client can break your model)

15. How do tech and ops work together?

We work in squads, pairing business owners (risk, sales, finance, etc.) with dedicated tech. Together, they act as the PM of the squad, it’s collaborative by design.

16. How do you test without breaking things?

LaunchDarkly is our best friend, powering smooth rollouts. We have massive test coverage, over 8,000 tests (unit, integration, E2E). It’s our biggest time investment, but it’s what keeps us fast and fearless.

17. What’s next for Defacto’s platform?

Make our platform fully self-serve and configurable so anyone, banks, financial institutions, fintech, SaaS, marketplace, can launch lending features in days, not months.

18. If you could rebuild Defacto today, what would you do differently?

I’d talk to more lending experts, not just fintech pros. The industry has nuance that takes years to learn, we could’ve accelerated some of those lessons.

19. What’s the role of AI in your stack going forward?

It’s everywhere:

• in the data processing to build a real time balance sheet based on banking transactions
• in our underwriting decision to analyse ad-hoc documents
• in our risk model to predict insolvency
• in our fraud detection
• in our recovery process to outreach delinquent borrowers
• in our support team
• in our dev team in cursor
• in our day-to-day, to build all the SQL queries to analyse your business

20. How do you manage tech debt in fast growth?

Never pause” business to refactor, but always know where your debt lives. When a project touches it, allow for 1–2 weeks of clean-up. Debt should get paid down over time, not ignored.

21. Advice to other CTOs building in B2B finance?

Fintech is 95% edge cases. The happy path is easy. Real life isn’t.

Invest in data modeling and accounting precision early. Losing track of money is a nightmare you don’t want to live.

Defacto’s journey shows how deeply technology and product are intertwined in the future of finance. By combining modular infrastructure, strong engineering principles, and a pragmatic approach to regulation, the team is building more than just instant lending. As Marco reminds us, scaling isn’t just about speed, but about ownership, precision, and building systems that others can trust!