Building Quantivo — From Idea to Seed Round

Building a fintech startup at 20 wasn’t exactly the plan. I was working part-time as a sysadmin in Pordenone, managing servers and legacy infrastructure, while studying Machine Learning and Big Data at the University of Udine. But when you see a gap in the market, independent financial advisors in Italy still drowning in spreadsheets and manual processes, you can’t unsee it.

The problem

Italy has thousands of independent financial advisors. Their daily workflow looks something like this: receive a PDF from a fund house via email, manually extract the data, open Excel, update the client’s portfolio, copy the numbers into another spreadsheet for reporting, and email the client a summary. Repeat fifty times.

The tools that exist are either enterprise platforms designed for large firms with large budgets, or glorified spreadsheets with a login page. There’s nothing modern, affordable, and actually built around how an independent advisor works day to day.

That’s the gap Quantivo fills.

Building it

We chose SvelteKit, FastAPI, and TimescaleDB as our core stack. Fast UI, Python’s ML ecosystem on the backend, and a database built for the time-series nature of financial data.

The feature that gets the strongest reaction from advisors is automatic document processing. They receive dozens of emails daily containing important data buried in PDFs and attachments. Our AI pipeline extracts, classifies, and matches that data to the right client portfolios automatically. What used to take an hour of manual work happens in seconds.

We built a feedback loop where advisors can correct mistakes, so the system gets smarter over time. That’s the part that makes them trust it.

What actually took the most time

It wasn’t the AI pipeline. It was the boring stuff.

Getting the permission model right, who can see which client’s portfolio, what can junior advisors edit vs. read only, took weeks before we wrote a single line of exciting code. Pricing took multiple iterations. Financial data sources are messy and inconsistent, and cleaning them up is unglamorous but essential work.

The lesson: the hard part of a startup isn’t building the cool feature. It’s building everything around it so the cool feature actually works in the real world.

What I learned

Talk to users before writing code. We interviewed financial advisors for weeks before building anything. That research killed features we were sure they’d want and surfaced problems we’d never considered.

Infrastructure experience helps more than you’d think. My background managing enterprise systems, databases, VPNs, DNS, the whole stack, turned out to be directly relevant when designing a cloud architecture for a SaaS product. Nothing in a startup is as stressful as a production database issue at 2am when you’ve already lived through it.

Type everything. TypeScript on the frontend, strict typing on the backend. The upfront cost pays for itself before you hit 10k lines of code.

What’s next

We closed our seed round and we’re focused on onboarding our first cohort of advisor firms. There’s a lot of work ahead, portfolio analytics, automated reporting, scaling the platform, but the foundation is solid and the problem is real.

If you’re building something, or just curious about the journey, reach out. I’m always happy to talk.