Working software, measured in weeks.

Affinitiv makes marketing and service software for automotive dealers. Its product line grew through acquisition. Moving those products onto one modern platform while dealers use them every day is slow work, and it had been underway for years. New products, meanwhile, often come with fixed outside deadlines.

We built two prototypes for new products, each in a four-week sprint against a fixed date, writing the spec first and letting AI do most of the typing. One is a service-department app covering scheduling, vehicle inspections and repair-order tracking. The other is a creative and asset hub with AI-assisted design tools. Both were built in real code, and Affinitiv adopted both as production software.

Underneath both sits the design system we built for the portfolio: a component library that works across the acquired products, with design tokens that hold the brand at the variable layer. It is why a new screen takes a day instead of several.

The engagement has run for several years and continues.

RVLVR runs through-channel marketing automation for large technology brands, including Intel, Apple, Cisco, HP, Ingram Micro, Lenovo, and Qualcomm. Forrester recognizes it in the category.

They needed to validate a new application surface before committing engineering time. The usual path is two to three months of discovery, wireframes, prototyping, and iteration. We did it in two weeks.

The prototype was real code: production-quality components with working state and interactions. It came out of a handful of focused interviews with RVLVR leadership and customers. Access was tight, so feedback loops closed in hours.

The project stays stealth. The engagement continues.

Sided is a polling app that turns polls into ad inventory. Adam is co-founder and COO. We list it as operated work: CAIO has no vendor contract with Sided.

Conventional planning put the core product rebuild at six to twelve months. It shipped in two, with the spec written first and AI handling most of the implementation. Sales work ran alongside it: personalized pitch pages for outbound, custom demos for buyers, the investor deck for the round, growth modeling, and category research on polls as ad inventory.

Sports Injury Central covers injury analysis and intel for sports betting. The hard deadline was the 2025–26 NFL season.

Same approach as Sided: write the spec first and let AI handle most of the implementation. We moved the platform off its old stack, built an algorithmic scoring model for injury signal, and launched the picks marketplace where bettors transact. All of it ran in parallel.

The engagement is multi-year and ongoing.

They had AI experiments running and nothing in production. The ask had two halves: a plan for where AI belonged in the business, and engineers who could keep building after we left.

The strategy work covered the roadmap, vendor and model selection, and a ranking of which workflow automations would pay off. The engineering work was hands-on. We paired with their team leads on production prototypes and ran training built around how the team already works.

ShedBuilder.ai is software for shed builders. Its parametric framing engine produces permit-ready drawing sets and cut lists. Charlotte Shed Builders and Lake Norman Handyman are working local businesses that run on it.

Every feature ships into a business with real customers and payroll first. When a workflow fails in the field, we hear about it the same day.

The local sites, charlotteshedbuilders.com and lakenormanhandyman.com, started pulling referral traffic from Google and ChatGPT soon after launch. Content and SEO run on shared multi-tenant infrastructure.