The thesis: ambitious consumer products belong in this studio too.
cloudgroup's core promise is that custom software is finally within reach — fixed-price, two-week sprints, full code ownership. The early proof points were operational: internal tools, admin systems, the kind of work non-technical founders had been priced out of for years.
But the bigger question loomed: does this approach work for an ambitious consumer product? Real-time data, polished mobile, social mechanics, a multi-platform launch — the kind of thing that used to require a venture round and twelve months. We wanted to know if the new toolset could compress that, too.
cloudbook was the answer to our own question. We picked a category we love (sports), a real frustration (most fans want the engagement of betting without the financial risk), and we built it the same way we build for clients: scope first, design second, ship fast.
“A two-week project that grew into the most-iterated thing we've ever shipped.”
Architecture: three layers, real markets in, reputation out.
The hardest design choice was getting the foundation right. Most fantasy products invent their own scoring; we wanted to anchor cloudbook in real sportsbook lines so that every prediction was honest. That meant building three layers, each doing one thing well:
The dashboard fuses the three layers — career stats, this week's performance, leagues, and your agents — into one home view.
What we built.
Four interlocking products, all in one platform:
- Cross-platform shellA Next.js web app and a native iOS app (Capacitor) sharing one codebase. Sign in once, place a bet on the train, check it again at home.
- All major sportsNBA, NCAAB, MLB, NHL, EPL, EFL Championship, MLS — live now. NFL and NCAAF light up when their seasons start. Spreads, totals, moneylines on every game.
- LeaguesPrivate or public groups with weekly bet requirements, league-scoped balances, ROI-based standings, and Sunday paydays. The social layer that turns a solo habit into a multiplayer game.
- AI agentsCustom betting agents — each user can deploy one with a written strategy. The platform fields six house agents that run 24/7, logging every decision (bet or pass) with reasoning.
Outcomes.
Beyond the numbers: cloudbook ships to production every couple of days. The same approach that ships a client's build in two weeks lets us push twenty improvements per month to our own product. Features that used to require a roadmap meeting — agent caching, ROI callouts, decision-journal scrolling — get spec'd, built, and deployed in an afternoon.
Economics.
A consumer product, built and launched, for the cost of a single Medium engagement.
cloudbook would have been a Medium engagement at cloudgroup pricing — $40K, four weeks, multi-feature product. We built it for ourselves to prove the math, then kept iterating because we like the product. The takeaway for clients: ambitious consumer ideas — the kind that used to need a seed round — fit inside a fixed-price engagement now.
How it felt.
“We built the product we wanted to use ourselves — and shipped a hundred versions of it before half our peers shipped one.”
— Mark Brogowicz, cloudgroup