The challenge
The founder had secured pre-seed interest from two investors and a term sheet meeting in eight weeks. The problem: the product existed only as a Figma prototype and a pitch deck. To close the round, he needed a working demo that could handle a live walkthrough — not a click-through mock, not a video, but a product that responded to real inputs in real time.
The idea was a cash-flow intelligence tool for small business owners: connect your bank accounts, and the product would surface patterns, anomalies, and projections in plain language rather than spreadsheets. The target user was a non-technical small business owner who was comfortable with banking but not with financial modelling. The product needed to be usable without a tutorial.
The constraint was real: six weeks from first conversation to something an investor could sit down with and use independently.
What we built
We ran a focused scoping session in the first two days to identify the smallest product that was genuinely compelling — not a feature-complete platform, but a coherent experience around the single insight that would make a small business owner say "I didn't know that, and I needed to."
That insight was cash-flow projection: given a month of transaction history, what does next month look like, and which of your regular commitments are at risk? Everything else — scenario modelling, multi-account comparison, export features — was out of scope for the MVP. Scope discipline was the primary engineering decision.
The data layer used Plaid for bank connection and transaction normalisation. An OpenAI-powered classification and projection layer categorised transactions, identified recurring patterns, and generated the plain-language summary that surfaced on the dashboard. We spent two weeks tuning the prompt structure and output format to get summaries that felt like a knowledgeable CFO, not a chatbot — specific, confident, and correctly hedged where the data was ambiguous.
The frontend was built in Next.js with Tailwind — fast to build, fast to load, and easy to iterate on between investor meetings when the founder needed to adjust the framing of a particular insight based on feedback. We built a demo mode that used realistic synthetic data so the founder could do a live walkthrough without needing to connect a real business account in the room.
Authentication, billing hooks, and a basic account management layer were included from the start — not because investors would stress-test them, but because a product that looks like it's ready to have real customers is more fundable than one that's clearly a prototype.
The outcome
The founder demoed the product in four investor meetings over two weeks. Two of those meetings extended significantly because the investors wanted to explore the product themselves. The round closed at $4.1M — meaningfully above the initial target — with one investor specifically citing the live product as the reason they moved quickly rather than asking for another meeting.
The six-week timeline held. The discipline that made it possible was the decision made in the first two days about what the MVP was not. Every feature request that came in during build was evaluated against a single question: does this make the projection insight more trustworthy or more useful, for the user sitting in front of it right now? Most requests didn't make the cut, and the product was better for it.
The codebase was structured for handover from the start — documented, tested at the critical paths, and deployed to Vercel with CI in place — so the founder could bring on an engineering hire post-round without inheriting a mess.
“I went into investor meetings with a live product that did exactly what I'd promised in the deck. Ryens built it in six weeks and made it look like we'd been working on it for a year.”
Got an idea? Let's ship it.
Tell us what you're building. We'll come back with a clear scope, a real timeline, and the senior team who'll actually build it.