Case Files
We close the tech void.
Then we show our work.
Client work stays confidential while it’s open. When a contract closes, we publish the whole thing: the problem, the build, what broke next, and how we closed it. No stock photos. No “synergy.” Just receipts.
Case 01Live
MOOD:ME: Teaching an Online Store to Let You Try It On
Every online clothing store has the same quiet problem: nobody can try anything on. MOOD:ME sells more than 500 made-to-order pieces in sizes S to 5XL, and the product photos showed one size on no one. So the shopper guesses, the box ships, and the return label does the talking.
The challenge
Shoppers buy with their eyes and return with their bodies. Plus-size shoppers had it worst: a flat lay tells you nothing about how a 3XL actually falls. The goal was simple to say and annoying to build: show the real piece, in the real size, on a real body, before checkout.
The build: a fitting room that lives in a browser
- 📷Your photoor pick a House Model
- 👗Build the outfitup to 3 pieces, color + size
- 📏True-size checkgarment vs. body, in inches
- ✨AI renderGrok image API, ~15 seconds
- 🪞Your lookwatermarked, to cart, share or The Salon
It runs on ordinary shared hosting, which has opinions about long requests, so every try-on is a background job the page politely checks on. Selfies are never stored. Results delete themselves in 24 hours. Every image carries a watermark, and the room is members-only, with daily limits and a spending cap so the AI bill never gets a vote.
The part most AI gets wrong
Image engines are people-pleasers. Ask one to put a 2XL hoodie on a size Small and it will quietly tailor the hoodie to flatter her. Lovely. Also fiction. So we stopped asking nicely. Before every render, the system compares the garment’s real chart measurement for the chosen size against the model’s actual body, then tells the engine exactly how it should hang.
Real bodies, on purpose
Fifteen House Models, Small to 4XL, each with real height, weight, measurements, body shape and a bio in their own voice. Pear, apple, rectangle, triangle, oval. Find your body twin, tap “Try a look on me,” and the Fitting Room opens with them already standing there.
Community, not just commerce
Members can send a look to The Salon, a public gallery, and every look is approved with one tap from an email. The Artist Palisade puts independent makers front and center, and a custom tee designer lets shoppers print their own art. A store people visit, not just one they buy from.
How we worked
Build the feature people can feel first. Ship in versions. Let the AI do the heavy lifting, but never let it guess the fit. Launch before perfect, and let customers tell you what’s wrong.
Case 02Closed
Twin City Produce: From Dot-Matrix to a 99 SQF Score
The warehouse already fed Northwest Arkansas. The office was still running on a freezing server, a dot-matrix invoice printer with strong opinions, paper timesheets and no domain. A food recall took four people and four hours.
The challenge
The week had outrun the tools. Accounting, time, product identity, the public front door and the credit file didn’t match the warehouse.
A freezing, failing server
→A server and QuickBooks the whole office can share
Invoices off a dying dot-matrix
→Invoicing that doesn’t require a prayer
Paper timesheets
→Time clocks straight to payroll, drivers and remote staff included
No domain, no company email
→twincitynwa.com, professional email, separate public, customer and team doors
A recall: four people, four hours, three desks
→Product Identification F2T: lots and vendors traced in versions
A specialty spice line that lived in one person’s head
→Spice distribution automation: a process, not a person
What we built
B2B distribution site. First domain, professional email and a wholesale store with 10 custom price lists by customer and customer group, so every buyer logs in to their own pricing. Public, customer and team access are split so the cooler isn’t run from a brochure.
The Team Portal. A complete company intranet: employee and hiring portals, the food safety program, business ethics, sustainability and workplace safety, plus Speak UP, a simple way for anyone to raise a concern. Safety and sustainability stopped being a binder on a shelf and became one click from every workstation.
AI operation systems. A shared server and QuickBooks, invoicing off the dot-matrix, workstations that survived the new box, and time clocks through to payroll. Food safety and quality shipped in versions: Product Identification F2T, the ECO ripening process and room, spice distribution automation and a security path for the floor.
Research credits & founder infrastructure. The projects were written up while they were happening: matrix, technical principles, labor and expense file, and documentation for the CPA. No refund number on this page.
The last SQF food safety audit came back a 99. The recall that once took four people and four hours doesn’t anymore.
How we worked
Start where the week is breaking. Ship in versions. Separate public, customer and team access. Write it down while it’s live. Leave a cadence, not a dependency.
Case 03Built 2023–2024Completed 2025
Firefighter Challenge League: The Store That Paid for the Platform
The Firefighter Challenge League has athletes, records, regions and a championship series fans travel for. What it didn’t have was a store. Or a single brand standard. Or the budget to build the independent platform it actually needed. So the brief was refreshingly blunt: launch the first official store, and make it sell enough to pay for what comes next.
The challenge
A nonprofit league with a global name and no storefront. Brand assets lived in a pile of print files sized for banners, not browsers. And the long-term goal, an independent, scalable web platform, needed capital the league didn’t have yet. The store wasn’t the finish line. It was the fundraiser.
A store shaped like the league
Competitors don’t shop like casual fans, and a Hall of Famer shouldn’t have to dig through keychains. So the site architecture mirrors how the league is actually organized: by region, by audience and by championship tier.
One brand, finally standardized
Before a single product went live, a Brand Mood Board locked the look: type, color and visual tone, so 300 products would read as one league instead of 300 opinions.
300+ products, designed in-house
As lead product apparel designer, we built the catalog end to end: custom competitor gear, official athlete flags and a full branded lifestyle line for fans. Gear made for people who run up towers in full turnout gear, and for everyone cheering them on.
Built on first-generation AI
This was 2023 and 2024. AI image tools couldn’t spell, couldn’t hold a logo still and had never heard of a size chart. So we learned to make them behave. When the generators couldn’t place a design on a garment correctly, the product mockups were hard-coded by hand. The UGC-style content and promo videos came out of the earliest AI tools, long before “AI marketing” was a LinkedIn headline. The workflow we sharpened here is the same one behind the MOOD:ME Fitting Room today.
Fast enough to keep the cart
The asset library arrived as high-resolution print files: perfect for a banner, brutal for a product page. Every graphic was reformatted and compressed for the web, cutting the rendering lag that quietly kills checkouts. Fast pages, fewer abandoned carts.
Marketing with a mission
Marketing reels and high-conversion promo graphics pointed athletes and fans straight at the new store. Then we stacked a partnership on top: the End of Year Fundfest, which routed proceeds from Nviroclean sales directly to the FCL to speed up their fundraising timeline. One client helping another reach the finish line.
- 🔥Brand lockedmood board, type, look
- 👕300+ productscompetitors, athletes, fans
- 🛒Store launchedfast, mapped to the league
- 📣Reels + Fundfesttraffic and partner proceeds
- 🏁Platform fundedindependent infrastructure
It worked. The storefront, the catalog and the campaign landed with athletes and fans and drove the sales volume the league needed. The First Responder Institute secured the capital to move to its own fully independent digital platform. Mission accomplished, and the store paid for its own successor.
How we worked
Lock the brand before the first product. Build the store the way the organization really works. Make it fast. Then market it like it matters, because the money is going somewhere that does.
Your turn.
Every engagement starts with a Tech Void Diagnostic: a 90-minute working session, a written memo and a 90-day plan for what to fix first and what to ignore. $1,250, credited in full to project work signed within 45 days.