15 years of enterprise design at USAA, Procore, and Document Crunch / Trimble. Currently operating RJ Digital — an AI-native agency where autonomous agents handle client research, copywriting, and community content, orchestrated by a Claude-based system I designed and built.
USAA's process for reporting the loss of a loved one was reactive — members had to call, re-tell their story to every representative, and navigate Social Security, the VA, insurance, and USAA separately, alone, while grieving. I led the team that rebuilt it as a digital compass: a single experience that walks surviving family members through every step, across every institution, on their own schedule.
Six weeks of discovery — listening in on rep calls, interviewing fifteen recently-bereaved members, co-designing with thirty-two workshop participants including five who had just lost a spouse. Six more weeks of prototyping and usability testing with twelve more bereaved members. The goal was 10% of all USAA death notifications through digital in two years. We hit 30% in year one.
Jane — take this at your own pace. We'll be here whenever you're ready.
Forge UX Patterns is a public library of 37 interaction patterns grounded in behavioral science — Fogg's behavior model, Kahneman's decision psychology, Wendel's design methodology, Norman's HCI principles — and explicitly adapted for AI agent product design. Every pattern carries a canonicity score, cognitive-load rating, and authority trace. Patterns like The Autonomy Dial and Strategic Friction give designers a shared vocabulary for the questions nobody was naming yet: when should an agent ask, when should it act, and how do we design for trust at every cognitive gate.
A visible interface control for how much autonomy a user grants an agent — from ask before every action to act-then-report. Reduces trust friction by making autonomy a first-class design decision rather than a hidden configuration.
"Most Principal candidates have worked near AI. I've shipped an agent system to production."
RJ Digital runs on Jarvis — a Claude-based orchestrator I designed that routes work across specialist agents: Atlas for SEO recon, Poe for copywriting, June for community content at Seguin Digital (our local publication where she publishes autonomously every morning), Librarian for file hygiene. Clients get agency-quality output at founder-scale overhead. The design work isn't the screens — it's the system: per-client brand context injection, cross-session memory, scheduled execution, and a pixel-art team room for status visibility.
On a February night in 2026 I built an auto-redlining prototype end-to-end — PDF extraction, GPT-4o risk analysis, Microsoft Word output with real OOXML tracked changes — in twelve hours, using AI agents orchestrated via a messaging app. I demoed it to CEO Josh Levy the next morning. He called it "the compelling event." The engineering head said the company "will die without agent-first development."
That prototype became policy. Document Crunch adopted agent-first prototyping as mandatory engineering practice, and the work grew into the AI Strike Team — where I architected the three-tier knowledge structure (Company, Industry, Local Market) that turns generic language models into domain-fluent AI employees, now shipped to real clients including TKG Custom Homes in Seguin.
The Contractor shall defend, indemnify and hold harmless the Owner from and against any and all claims arising out of resulting from the Contractor's negligent acts or omissions in the performance of the Work...
AI is a material, not a layer.
The designers who bolt AI onto existing products ship things that feel uncanny. The ones who treat AI as a material — something with grain, with constraints, with moments where it bends beautifully and moments where it cracks — ship things that feel right. Start from what the model can actually do, and design with it, not in front of it.
The design work isn't the screens. It's the system.
Screens are the output. The system is what decides whether the output is good. Tokens, context injection, state management, scheduled execution, error recovery — that's where the leverage lives. If your design system is load-bearing, the screens almost design themselves. If it isn't, no number of Figma files will fix it.
A moment of pause is research.
Usability testing at USAA taught me that waiting is a skill. When a member paused on a screen and I didn't rush to ask the next question, they gave me a better answer than any prompt would have pulled out of them. Don't resolve silence. Don't clarify ambiguity. Let the person bring themselves to the problem — you'll design for what they actually need.
Speed changes strategy.
When a prototype takes three sprints, it's a commitment. When it takes twelve hours, it's an experiment. Agent-first development doesn't just save time — it changes which decisions get made, because leadership can test instincts instead of planning around them. The fastest way to find out whether a thing is worth building is to build it.
I'm Ryan. Fifteen years of enterprise design, based in Seguin, Texas. I design and ship AI-native products — the most recent one is the agency you're looking at.