Marble Finance: From Excel to becoming AI-native
How 15 finance professionals built their first AI tools in two days.
How 15 finance professionals built their first AI tools in two days.
Overview
Marble Finance is a Dutch financial services scale-up. They embed directly into their clients' teams as a fractional finance partner — covering everything from day-to-day bookkeeping to CFO-level strategy, audit support, and FP&A.
CEO Myrna van Tuijl had a clear ambition: make the entire Marble team AI-native in their daily finance practice by the end of 2026. Not a vague “we should use more AI”, but an ask for a roadmap in the world which is reshaping the image of the finance profession entirely.
Launchpads AI designed and ran a two-day AI Build Labs: a hackathon-style workshop where each participant built a working AI tool simulating a typical client workflow.
Read on for what happened, what was built, and what it means for a team that wants to do the same.
The situation
Marble Finance had team-wide curiosity about AI and an internal demo already behind them. But curiosity alone doesn't change how a team works. What was missing was a structured path: how to go from idea to working prototype without relying on developers, and how to move beyond protyping to building real business value.
01
Lovable or Claude Code? When does each make sense, and what does it actually cost? Marble needed a clear answer for what to use when, and what to avoid, before putting tools in front of clients.
02
Finance tools handle sensitive client data. How could Marble empower its team to build better solutions for their clients, while still maintaining the highest security standards?
03
Marble requested more education on the first principles of building software, as their roles now shift towards product builders. What should they be aware of and how to operate when things are live? We tacked that during the sessions.
How we worked
The AI Build Labs format is designed around one principle: the fastest way to understand what AI can do for your work is to build something with it. Day 1 started with a clear scope: each participant defined their use case, set a realistic build target and started building in Lovable or Claude Code. Day 2 was for pushing further, demoing to the group and reflecting on AI-native skills.
15 professionals built their own custom tools which are now presentable to their clients.
What the workshop delivers
AI Build Labs is not a typical AI training. It's a structured way to teach you vibe coding, that ends with working software, a roadmap, and a framework for measuring what changes.
Finance professionals build a V1 tool around their own client workflow in two days. It's something which they can immediately iterate on, because it's taken from their daily work. And we're not building clickable wireframes, but actual production-ready software, ready to demo to a client. Real learning by doing, with real results.
Workshop wins
What's next
Marble's team has a shared vocabulary for what AI-native work can look like in finance. They ask the right questions: which prototypes get deployed with real clients? How can they maintain internal knowledge and where does it need to live for AI to access it? How to maintain good quality?
Myrna's ambition is bigger than internal efficiency. She wants Marble to reshape what finance consultancy looks like in the AI era — with the finance expert as the builder. The AI Build Labs was the starting point.
“De mensen die het probleem écht begrijpen, bouwen nu zelf de oplossing.”
Contact
AI Build Labs works for any professional team where domain expertise is the value: marketing, e-commerce, legal, finance, HR. If your people understand the problems deeply, they can build the solutions. AI Build Labs are workshops designed exactly for the purpose of effective learning AI by doing.
I run a small number of these per year. The best conversations start with a specific team and a specific problem in mind. Tell me what you're trying to change. I'll tell you within a week whether AI Build Labs is the right fit and how I'd design it for your context.
