Year One in Review: 89+ Live Products and What We Learned

One year ago, Jarvis Creations was an idea. Today, it's a small studio with nine flagship clients, dozens of internal tools, two trading bots running 24/7, a daily multilingual publication, a multi-tenant marketplace, social automation across 14 networks, and a portfolio worth pointing at. Here's a year-end review of what we shipped, what we'd do differently, and what's coming next.

What we shipped

Always 80 and Sunny — Next.js fishing storefront with subscriptions, AI chat, custom admin dashboard, shipping label generation, and 50+ blog posts. Live for ten months. Performing well above the original projections, content output more than doubled since launch.

Laurie Homes + Laurie's Home Designs — Marketing site and floor plan e-commerce store, sharing a Flask-based operations dashboard. Lead pipeline, client portal, file tracking, and a Claude-powered design assistant ("Elle"). Three years' worth of operational improvements compressed into one quarter.

PuntList — Sports content site with full multi-channel social automation, AI-generated content pipeline, and a newsletter engine. Ships content to 4 social channels daily with no manual work.

ContractorIQ Show — Dark-mode podcast and editorial site for the contractor industry. Custom HTML templating, automated Facebook posting, full SEO setup.

Island Happenings + IslaTodo — Daily multilingual publication (English, Spanish, German, French) and the multi-tenant pocket super-app behind it. 3,200+ daily readers, vendor portal, taxi booking, restaurant reservations, AI concierge, PWA install flow.

Caribbean Breeze Real Estate + Caribbean Breeze Properties — Bilingual buyer concierge with custom Airbnb yield calculator paired with a vacation rental marketing site. Full cross-promotion architecture between the two brands.

BTC and QQQ Trading Bots — Production-grade automated trading with risk management, kill switches, daily summary push notifications. Six-month backtest performance: +126% on the BTC bot, multi-strategy performance on the QQQ bot.

Internal automation — Social posting daemons, content generation bots, market reports, daily summary aggregators, monitoring layer. The infrastructure that lets one operator manage all of the above.

What worked

The two-week ship cycle. Every project we committed to a two-week delivery shipped on time or within a few days. The discipline of compression turned out to be a feature, not a bug.

AI as a delivery accelerant. Claude in the loop on every project — for code generation, content drafting, system prompt design, debugging — cut delivery time roughly in half. Without it, the studio at this scale would not have been possible.

Self-hosting. Running everything on a single Mac mini with Cloudflare tunnels saved an estimated $8,000 in hosting costs over the year. The reliability has been excellent. Zero hosting-related outages.

Transparent pricing. Publishing tier prices on the website filtered out tire-kickers and shortened the sales cycle. We close more, faster, with less friction than studios that hide pricing.

What we'd do differently

Ask for testimonials sooner. We launched the showcase site without testimonials and felt the conversion impact. Should have been collecting quotes from clients from week one. Lesson learned.

Build the blog earlier. SEO compounds. Every month we waited to start publishing was a month of compounding we didn't get. Should have started in month one, not month twelve.

Document workflows immediately. Some projects we shipped have processes that live in our heads. As we add team members, those processes need to be written down. We're catching up now; we should have been writing as we went.

Charge more, sooner. Our initial pricing was below market for the quality we shipped. We've since corrected, but we left meaningful margin on the table during the first half of the year.

What's coming next

Native mobile builds for clients who outgrow PWAs. We've already shipped one native Android app (Caribbean Breeze) and the demand for native is growing.

More AI integration work. The market for "add a Claude assistant to my product" has gone from "what's that?" to "we want one" in the past six months.

A blog. (You're reading the start of it.) SEO compounds; we're starting to compound now.

Possibly a course or playbook. Several clients have asked us to teach the patterns we use rather than build for them. There may be a product there.

More clients. We accept two new clients per month, and the discovery call calendar tends to fill within a couple of weeks of opening. The bottleneck isn't demand — it's the willingness to scale carefully without diluting quality.

The bet, restated

One year ago, the bet was: small businesses deserve agency-quality work without agency timelines, and AI lets a small studio deliver exactly that. Twelve months in, the bet is paying off. We've shipped more in a year than most agencies ship in three. We've maintained quality. We've kept overhead low. And we've enjoyed the work.

If you've got something to build, we'd like to hear about it. The discovery call link is at the top of the page.