One builder ships what five used to.

The situation

I run the design org at Conviva: UX, research, and content, distributed across two continents. The team worked the way most enterprise design teams work. Designers made mockups in Figma. Engineers interpreted them. The distance between the two was measured in spec documents, review meetings, and rework.

That pipeline was built for a slower company than the one we were becoming. Product timelines were compressing while design headcount was not. We were running at 105% capacity across multiple P0 projects at once. I partner with product and operations on design's budget and resource allocation, so I knew the honest version of the constraint before anyone said it out loud: more designers were not coming. The output per designer had to change, not the number of designers.

The reframe

Everyone around me saw AI as a way to make the existing pipeline faster. Figma plugins, quicker wireframes, the same artifacts produced sooner. I saw the artifact itself as the problem. A mockup is a promise that someone else has to keep, and every promise costs interpretation, meetings, and drift. If designers could ship working code, the handoff would not need accelerating. It could be removed. I had already made this argument in How to Lead Design in the AI Era. This was the chance to prove it in production, with my own org.

The work

In late 2025, I moved the team off Figma and onto an AI-augmented prototyping workflow built on Bolt.new. Not a pilot, not one squad, the org. Killing the safe Figma-to-spec-to-ticket pipeline in the middle of the highest-stakes project of the year was a bet nobody asked me to make. The AI coding tool had barely existed six months earlier, and no design team I knew of had adopted it at this scale. I made the call and I owned it. Designers now build functional prototypes, and engineers use those prototypes as reference implementations. The oldest question in the pipeline, what did the designer mean, mostly disappeared. The answer runs in a browser.

The mechanic is simple: designers work in the real production system, not a picture of it. They iterate against live components and hand back working software, not static screens, so engineering inherits something that already behaves instead of something to interpret. For early discovery I built a prompt-to-prototype model on top of this: working, high-fidelity prototypes at conversation speed, before anything earns a roadmap commitment. Discovery that used to yield three wireframes in a month now yields three working prototypes in a week. The pod that ran that discovery, several people meeting over weeks, collapsed to one designer and one PM.

A workflow this fast needed a governor. So on my own time I built Bolt Buddy, an AI agent that sits between a designer's intent and the coding tool. Before a prompt reaches the tool, Bolt Buddy rewrites it to comply with the design system and WCAG 2.2 AA: a request for a generic green comes back as the brand green, an off-spec radius gets corrected, an attempt to alter a shared component is blocked. It catches the drift before it becomes code. Bolt Buddy governs the how. I govern the what and why. That split, strategic decisions to the human and execution guardrails to the agent, is what AI-governed design operations actually looks like.

I ran a from-scratch design-system remediation across the codebase in parallel: standardized components, a consistent neutral token palette, and the brand green as the primary interactive color. The philosophy was to ship clean rather than ship noisy. If a feature was incomplete, I removed it rather than deliver confusion.

Language got the same treatment as pixels. I directed the UX writing function and built LLM-powered tone and microcopy frameworks that now govern content across product, docs, and marketing. One voice, encoded once, enforced everywhere, instead of a style guide nobody opens.

The hard part was not tooling. It was judgment. A designer who ships working code owns decisions a mockup lets them defer: states, edge cases, real data behavior. Leading that shift across two continents meant raising the standard of what counts as done, not just teaching people new prompts.

The outcome

The handoff pipeline compressed considerably. Design intent now survives contact with the build, because engineers work from prototypes that already behave correctly rather than pictures that suggest behavior. The prototyping model returned three high-fidelity prototypes in its first week, work that used to take a full team much longer. Bolt Buddy made the speed safe: design-system and accessibility compliance moved from something reviewers catch after the fact to something the agent enforces before a line of code is written. The tone and microcopy frameworks govern content across product, docs, and marketing, so the product sounds like one company. And the org that does all of this is the same globally distributed org I started with: UX, research, and content. Nobody was added. The deliverable changed, and the capacity followed.