Hey Xero 👋
Summary
I’m Buzz. I love designing in the unknown — the messy, stressful, uncertain space between a problem and something amazing. But I’m equally excited by creating the conditions that help designers navigate it and produce world-class work. I’ve spent more than 20 years scaling that creative capability without lowering the quality bar.
I spent much of my early career believing process was the enemy of creativity… rulebooks, systems and documentation were, in my mind, just ways of boxing designers into rigid, boring and repeatable structures. Call me a design hippie, but I value creativity and imagination above all else.
What I eventually realised was that the best creative process is often invisible — embedded in culture, collaboration and a handful of forcing functions that make the work better. Alongside the time I spend on the tools, I’ve spent the second half of my career designing and refining those conditions. Now, as AI expands what designers are capable of, this work has become both a new obsession and a growing urgency.
“Buzz cares so deeply for the customer, works so consistently to improve his craft, and makes every person around him better at their job.”
Nick Francis — Former CEO @ Help Scout
This is the story of how, over twenty-one years, I’ve helped organisations scale their creative practice… and how those lessons now shape the tools and processes helping design teams thrive with AI.
No time for a good story? Jump straight to the AI tools part.
It starts with teamsDirect link
Having hired and worked with so many brilliant designers, I’ve come to realise that individual talent is only the beginning. The real magic comes from building teams that obsess over their craft and push one another towards truly exceptional outcomes.
Great teams combine complementary skills and diverse personalities, optimise for trust and leave room for people to grow. These are the teams I’ve built and the ones who are able to adapt to change without compromising quality. Whether using AI to push a subtle animation to production simply because we’ve always wanted to, or sweating the meaning behind an internal brand project — people who challenge one another produce work that keeps getting better.
At Campaign Monitor (above), I built an in-house studio that designers loved working in and where they could produce some of their best work. It was modelled on the agency environments I’d previously worked within — messy, fun and a little chaotic, but organised around the work in progress. It was a physical expression of the best collaborative practices — brand designers intentionally sat alongside front-end engineers and influenced the product, product designers critiqued brand work and everyone shared unfinished ideas instead of waiting for perfection.
I stayed close enough to the work to unblock a project or demonstrate the level of quality I expected, but it was here that I learned my more lasting contribution was helping designers develop their own answer. After years of honing a collective creative voice, the team’s judgement had become transferable enough to survive beyond me.
“Buzz was happy to roll up his sleeves and help solve problems, but also understood when it was important to guide others on the team and teach them to tackle design challenges.”
Dave Greiner — Co-Founder @ Campaign Monitor
Nine years later, at Help Scout — a globally remote company — I applied the same idea without directly managing the designers producing the work. My role was to lead through example, create the conditions for quality and measure my success by theirs. As Principal Designer, I embedded myself across workstreams — helping teams frame problems and set more ambitious appetites, then moving between critique, coaching, prototyping and direct contribution according to what the work needed.
The intention was never to become a bottleneck. It was to give designers access to my experience and perspective so they could make bigger, more confident bets without taking authorship away from them — the same principle that guides my coaching practice.
Prevue was this same belief expressed as my own product. It was built around the idea that creative work becomes better when people can see and discuss it while it is still being made. Prevue celebrated exploration, abandoned directions and the uncertainty of the messy middle. Collaboration was not approval at the end of the process — it was part of the creative act itself.
My excitement about the work is rarely subtle. My aim is to make it contagious — giving designers permission to care deeply and raising the team’s appetite for what might be possible.
Then comes processDirect link
Summary
At the beginning of genuinely interesting work, nobody really knows what they’re doing — and I call BS on anyone who claims they can perfectly scope their way to an amazing outcome. That’s where good design process comes in — not as a formula for the answer, but as a framework for finding one. It means I don’t always need to know exactly how to solve a problem to feel confident that we’ll find our way to something exceptional.
But even the best process requires the right organisational conditions. The wider business needs to understand the value of design, tolerate some ambiguity and give teams enough room to explore. That permission has to be earned.
I earn it by making design less mysterious… embedding with engineers, working alongside Support, understanding the commercial expectations executives are balancing and making the investment in the unknown feel less risky. By establishing repeatable practices, educating teams and never allowing design to be a bottleneck, I earn the right to move deliberately when it matters.
At Buildkite — a deeply technical company with a natural bias towards code — complexity was familiar and celebrated, while design remained a mysterious skill that wasn’t widely understood. Reporting directly to the CEO and working closely with Support, Sales and Engineering, I helped make design visible as a collaborative function tied directly to customer success. Through extensive documentation, workshops and public events — alongside a design career ladder and new collaborative practices — I helped make design a funded and celebrated capability across the business.
Fun Fact…
Shortly after ChatGPT launched, I dedicated a week with the Buildkite design team to exploring ChatGPT and Midjourney. That early experiment eventually led to me launching 2 Weeks AI — a crash-course in AI that has since been completed by 50,000+ people.
Above A fraction of the documentation I wrote to establish design as a function
At Atlassian, the challenge was bringing creative practices to a much larger organisation without reducing them to another rigid playbook. My mandate when joining was simply to “improve creativity”… the designers had access to a world-class design system, but the system alone didn’t automatically translate into world-class outcomes.
Ironically for a technology company, much of my approach involved taking design back to basics. I introduced lightweight practices around sketching, writing and collecting inspiration, then worked with leadership and within the established structures to challenge where research and playbooks supported creativity.
I even created a physical space where designers could collaborate in the open and leave unfinished thinking visible (sound familiar?). It was here I started leaning into how creative practices, especially ones centred around discovery, could scale in a remote environment.
At Help Scout, the challenge changed again. Without a shared physical studio, I had to recreate the conditions that made those environments work across borders and time zones. Collaboration could no longer happen by accident… it had to be designed.
Through years of actually doing the work, I introduced lightweight asynchronous practices for sharing early and often. The goal wasn’t more process — it was preserving the vulnerability, spontaneity and context that physical studios provide for free. Remote collaboration has since become a defining part of my practice, and something I’ve written and spoken about extensively.
Through my advisory practice, I see these same patterns recur across some of the world’s strongest product teams. This creates a valuable feedback loop between my work at Help Scout and my coaching, writing and speaking… what I learn in one environment informs how I help another. That broader perspective has helped me remain in the unknown for longer and build repeatable practices that support creativity without pretending to manufacture it.
“Buzz has the rare ability to provide designers and design leaders both actionable, practical advice as well as high-level, strategic direction. His mentorship has elevated me and my team’s efficiency and performance dramatically.”
Tina Hsu — Group Lead @ Canva
Systems for scaling qualityDirect link
Summary
I was originally against design systems. Copy/paste solutions promising to scale design had always seemed to flatten creative work into beige sameness. On the other hand, bespoke solutions were slow to create, difficult to maintain and impossible to scale. In 2016, I reluctantly began working seriously with design systems, promising myself I would balance both… speed without compromising quality.
I didn’t know it then, but learning to ensure quality without constraining creativity would become the foundation for AI enablement a decade later.
At Atlassian, I worked to bridge the gap between central Design Ops and the product designers using the system. Mostly, this involved coaching designers to break the rules deliberately — then creating processes through which useful exceptions could improve the system rather than simply break it. At Buildkite, I created the company’s first design system around the same ideas of consistency, adaptability and shared ownership. It gave engineers greater design powers, allowing them to produce their own solutions while continuing to meet a shared visual and accessibility bar.
Along the way, I learned that the secret to a design system that stands the test of time is a tolerance for change. That was one of the core principles behind HSDS, Help Scout’s design system, which I created in 2018. It has weathered the better part of a decade and multiple generations of the product because it was designed to evolve rather than resist change. I expand on these points in this Design Ops Webinar.
Even with all that process and all those guardrails, I’m always mindful not to build so much structure that play and creative expression disappear. In all my design systems work you’ll see this dedication to freedom.
Above Making data interfaces fun for FairSupply
That tolerance for change is now proving essential. In early 2026, I began rebuilding HSDS to be legible not only to designers and engineers, but to our AI teammates.
Building tools to design with AIDirect link
In 2026, that foundation suddenly became essential, suddenly — designers across the industry were asked to move radically faster without allowing speed to erode quality. That demanded more than sprinkling a little ChatGPT over the existing process.
I led this shift on several fronts at Help Scout. I brought in external specialists for hands-on AI prototyping workshops and training in Cursor and code tools. I researched emerging tools and workflows, then codified what I learned into practical guidance for the wider organisation (read my writing). I also worked with leadership to define where AI could compress execution time — and where ambiguity still required deliberate human involvement.
My approach was to move fastest where decisions were cheap to reverse, then reinvest those gains in the moments where discovery needed time. Here are some of the ways I helped the team increase speed without compromising quality:
Encoded design judgement
I wrote Help Scout’s AI operating principles and design principles, then distilled them into a design.md — making the largely unspoken parts of our 15-year-old product philosophy and visual language explicit enough to guide both people and machines.
I came to appreciate just how important this layer of articulated judgement was, and that AI struggled hard when asked to create it using one of these tempting generation tools. It turns out even our robot overlords couldn’t infer years of accumulated taste, context and intent — so most of the work had to be done manually, over endless coffees.
Rebuilt HSDS for AI
I rebuilt our entire design system to make it legible to LLMs, restructuring more than 200 primitives and compositions, standardising hundreds of tokens and variables and writing the accompanying documentation layer.
Machine legibility also required complete parity between Figma and React. Without it, AI was confidently producing interfaces that appeared correct but were coded from invented components… essentially defeating the point of the system. I audited the React library, connected components through Code Connect and created or deprecated anything that no longer matched — running evals along the way.
I then used the restructured system and AI coding tools to build the missing React components myself, clearing work that had remained in the backlog for years. This became a practical way to test the system, expose gaps and improve the quality of its documentation.
Finally, I created reusable AI skills (above) that enabled Claude to create and maintain Figma-React documentation to the same standard — turning work that previously required significant designer time into part of the system itself.
Built tools for speed
I created Scenario Pal, a custom GPT trained on our company strategy, personas and customer analytics. Designers could use it to explore how concepts might perform across realistic customer scenarios and industries, or review screens and workflows against our brand, accessibility standards and design.md guidance.
I also partnered with engineers to create HSDS Studio (above), an internal AI prototyping environment grounded in our actual design system. It can generate complete Help Scout experiences from a Figma file or prompt using real HSDS primitives. It’s similar to Claude Design, but because it has access to our full design system, experiences are produced using production-ready code, verified components and created to our visual standards. From here code can be easily graduated to production.
Proving it in production
Once the infrastructure was in place, I challenged every designer to ship one change directly to production — and went first. Each designer followed, with their contributions gradually increasing in complexity and depth.
Within the first three months, I personally shipped more production code than I had in the previous decade. I cleared longstanding backlog items, added details that had repeatedly lost out to larger priorities and partnered directly with engineers to accelerate more substantial features.
Above I was able to take a fun and frequently requested little resize feature to the product in under an hour. Below Taking a fully functional table re-design from imagination to reality using all existing components.
Designers are now coding components, stress-testing ideas through functional prototypes and shipping selected improvements themselves. They can run local environments, share more of the QA burden and pair directly with engineers on work in progress. The result is more than faster output. The distance between designing and building has narrowed, long-neglected details are reaching customers faster and the team has greater capacity to pursue more ambitious ideas historically cut during scoping.
And true to my word, design has not become a bottleneck. We’re shipping more than ever, with greater confidence and excitement, while protecting time for the bigger problems — the ones where quality depends on dwelling in uncertainty.
Still designing in the unknown, still refusing to lower the quality bar — just bringing AI along for the ride. 🤖✌️