How we helped teams at Coinbase optimise their product development processes using Figma and AI technology.
Confidentiality notice
Introduction
As product teams in tech companies look for ways to deliver faster and more efficiently, AI technology creates new opportunities to streamline workflows, reduce repetitive tasks, and improve quality of output.
Context
Coinbase partnered with Skippr — a small start-up building AI products, with the goal of optimising their product delivery process. The traditional method of designing and building products is generally quite inefficient, with tasks that are repetitive, time-consuming, and tedious. As the company streamlines their work, our partnership sought to find opportunities in the realms of design, copy and code.
My ROLE
A primary contributor to the project on behalf of the AI start-up Skippr — leading design and user research, and involved in product strategy, QA, testing, and non-technical documentation.
Additionally I shared the duty of product manager with our CTO – writing requirement documents, liaising with engineers, defining project roadmaps, and planning product strategies and experiments.
Project goals
Automate repetitive tasks
Parts of the design, copywriting, and coding workflows involve repetition that eats up valuable time.
Reduce friction
in workflows
Product team contributors encounter barriers and frustrating points of friction in most tasks which hinder productivity and problem solving.
Improve
design system compliance
As a design function scales it becomes harder to maintain and adhere to design systems and product guidelines.
Improve consistency in products
As companies grow, cases of human error inevitable increase and slip through the cracks.
We needed to understand and address the day-to-day workflows and needs of the main contributors involved in building and delivering product solutions. While product teams generally follow best practices across different companies, there are always nuances and important details to account for.
Product designer
Concerned with solving core product and business problems – copywriting, developer handoff, design system compliance, etc are secondary
“I spend too much time on housekeeping instead of solving problems”
Design system manager
Needs to ensure the design system is adhered to, and designers are aware of updates or changes to patterns and components
“There’s always some level of compliance issue – using outdated components or colour styles for example”
Content manager
Copy should be ready for production faster, with less manual intervention – this becomes increasingly difficult as the product grows
“We have strict rules about how we describe certain terms and policies that change often, it’s really hard to enforce these”
Front-end engineer
Goal is to build products that are robust and reliable. Often receives designs that aren’t properly documented, leading to ambiguities and issues later down the road
“Many designs we receive aren’t ready for development yet”
The common thread across teams, and place where the work of our main users intersects, is Figma. More than just a design tool, it's where projects are reviewed and discussed, copy is written and edited, and important decisions are made. In this case a Figma plugin would serve as a way to access the workflow automation tools directly.
A Figma plugin bridges the gap between AI capabilities and day-to-day workflows. We kept the following guidelines in mind when designing the experience.
Unobtrusive
To maintain focus on the canvas and its contents, the plugin should be as compact as possible
Contextual
Consider the current task of the user and show corresponding info or actions accordingly
Consistent
All plugin features should work in a similar way and new additions should be familiar to use
Readily available
Provide quick access to relevant tools as needed, avoiding context switching
When starting a new task, designers often get frustrated by trying to find the relevant materials. In many cases a designer, a) doesn't know which design asset to use for a task, and b) where to find the correct versions of those assets. Existing flows or patterns will be used as references, meaning they'd need to sift through design files or reach out to colleagues to (hopefully) find what they need.
The Search feature of the plugin provides users with direct links to curated screens, flows, and components so they can find what they need quickly and move on.
Drawing from official Coinbase content and brand guidelines, production-ready copy from the CMS, and trained LLMs — designers and copywriters generate or fetch existing copy straight into their layouts.
Designers and writers can spend hours articulating and rewriting copy, trimming length and refining tone-of-voice. The challenge of communicating effectively within limited space is trivialised with the plugin.
Every designer (and developer) has been there — the solution is handed over for development and everything looks good at a surface level. However upon closer inspection there are outdated or detached components, mismatched colour and typography styles, or assets no longer supported. This leads to tedious design revision and maintenance, and even unnecessary rebuilding in code.
The plugin analyses the targeted screen or component and compares its contents with the design system, identifying and resolving issues in a fraction of the usual time.
Stakeholders, admins and design system managers need a place to monitor the efficacy of the system — which teams are making use of the tools, what parts need more work, and what usage trends are starting to emerge. The dashboard provides an overview into key metrics to make decisions about where to focus next.
The component matching engine analyses design system components and compares them to their coded counterparts. This means that when code for a new screen is generated, the correct code component is referenced — taking the guesswork and manual configuration out of the picture. The platform allows managers to ensure the matching is accurate and rectify any discrepancies.
Important design assets, content and brand guideline documents, and other sources-of-truth are managed in library. Good maintenance of this library is crucial to ensure end users get accurate results from the plugin in Figma.
The pilot showed legitimate potential and adoption for Coinbase
While the project is ongoing, the pilot facilitated significant buy-in from all levels including the CEO and EVP of Engineering. Naturally there are many obstacles to still overcome, but a viable foundation has been laid for the next stages.
While the project is ongoing, the pilot facilitated significant buy-in from all levels including the CEO and EVP of Engineering. Naturally there are many obstacles to still overcome, but a viable foundation has been laid for the next stages.
Thanks for your time! I'd be happy to elaborate on the work described in this case study on a call.
Ben Brouckaert Design + Product Porfolio
2025