Episode 075 Raw Data

Thanks for your interest in The Question!

Sign up to participate in future questions here.

This file is the raw data from Episode 075, deep dive hosted on May 21, 2026 with Ben Callahan and Cassie Groos.

There are 60 answers. The question this week was:

-----

Hello, system thinkers!

Last week, Cassie Groos and I were discussing something that's been nagging at a lot of us: design systems are having an identity crisis.

In our conversation, Cassie said, "If it's for humans then it's documentation, if it's for AI it's context." This sentence stuck with me and it felt like a concise way of explaining how our roles are evolving. The assets we ship, who (or what) consumes them, and our day-to-day design system work all feel somewhat different than they did a year ago.

Some of what's on our plates is brand new: context files, agents and skills, components as data, guidance on how to use AI with the system… And some of what used to be on our plates is shifting. The grunt work in Figma, component builds, even chunks of documentation. The role of a design system is clearly in flux (as are all the roles in our industry) yet we’ve felt protected from that for a while. There's a comforting story design systems folks have been telling ourselves, that AI needs us, that good systems are the moat. Cassie and I believe that narrative needs to be critically evaluated.

This week, we want to talk about the design system identity crisis.

Welcome to Episode 075 of The Question, with Ben Callahan and Cassie Groos!

--

How would you describe your design system team's posture toward AI right now? (Select one)
Leaning in hard
Experimenting carefully on the side
Watching closely but not changing much
Skeptical, the hype outpaces the reality
N/A or Other (please explain)

There's a common narrative that design systems are AI's necessary foundation. How much do you believe that protects your role as a design system practitioner? (Select one)
Strongly (it’s a big advantage for us)
Somewhat (for now, but not for long)
Not really (it's a comforting story)
I hadn't thought about it that way
N/A or Other (please explain)

Describe how your role as a design system practitioner has changed in the last 12 months. (This doesn’t have to be about AI!)

If you had to redefine "design system" in one sentence for someone joining the field today, what would you say?

What's a bet you're making about design systems right now that you might be wrong about by this time next year?

Question 1: How would you describe your design system team's posture toward AI right now? (Select one)

Question 2: There's a common narrative that design systems are AI's necessary foundation. How much do you believe that protects your role as a design system practitioner? (Select one)

Question 3: Describe how your role as a design system practitioner has changed in the last 12 months. (This doesn’t have to be about AI!)

Question 4: If you had to redefine "design system" in one sentence for someone joining the field today, what would you say?

Question 5: What's a bet you're making about design systems right now that you might be wrong about by this time next year?

Leaning in hard

Strongly (it's a big advantage for us)

Ive come a lot more visible, interest in contributing has increased. Working with Engineering a lot more and closer

a system that governs how experiences evolve, not just how interfaces look.

the need for Figma to design a component

Experimenting carefully on the side

Somewhat (for now, but not for long)

More focus on (non-conflicting) documentation, alongside working on more advanced component features while our enterprise design system matures.

Speeding up teams in creating interfaces with a good design, and connecting people in the process.

That more and more teams will be vibe-coding, or vibe-desiging in a enterprise organization. It might turn out that the disadvantages weight heavier than the advantages.

Experimenting carefully on the side

Somewhat (for now, but not for long)

As a designer and developer, I've seen a much bigger shift in the latter. I rarely write code these days and when I do its a couple of lines that's faster to write than argue with an agent about. But as a designer, by role feels largely similar, but I do find myself defending the need to use existing patterns and approaches more than before, as more people feel enabled to 'try out' designing things.

A design system is a set of rules, guidance and patterns that governs how we build things to fit our brand.

(It can include components, tokens, ect, but that no longer feels like the primary output).

Great interfaces, components, brands, ect will still primarily be designed by people. But not every team wants or needs to be great.

Our design philosophy and way of working has always been technology agnostic. So like any other tools we have used (in the past) to help create or express a design language (i.e. the design principles that you base high level design decisions around), it doesn't fundamentally change or replace how we go about creating user experiences.

For us, the difference between being able to create a design language and a design system is the difference between inventing a language and writing a book in a particular language (or even multiple languages).

But that might change once AI tools are powerful enough to create new languages...

Not really (it's a comforting story)

I think design system practitioners are having to ask themselves whether they've done enough to lift the maturity of the design community from simply churning out more artefacts (and how to do it cheaper/faster/better), or if we will see true integration between research, design, prototyping and testing with future design systems now that we can let AI tools do the heavy lifting with creating the artefacts.

I feel like my role as a design system practitioner now is to educate and help new designers build up their soft skills, and the most important ones are: curiosity, empathy and humility - things that we can't rely on AI to teach us (at the moment).

A design system is the implementation of a design language (i.e. principles and objectives of a design philosophy) so that it can be applied consistently to create reproducible and extensible interfaces.

It is possible that design systems will not be created by UX designers exclusively (or anymore), because there will not be the same distinction between designers and developers based on the old concepts about product design work means.

My company does not have a dedicated design system team, however there is top down leadership demand for the design system to lean hard into AI. This is adding excessive pressure to designers who split their effort between their product work and design system maintenance/management to learn and build fast (often then breaking things).

Strongly (it's a big advantage for us)

The AI context has given Design leadership a new point of argument to present to stakeholders and other company leadership for having dedicated design system team resources.

I like to think of it in terms of Lego. The design system contains the building bricks, some individual small pieces (e.g. button) and some some components already built out (e.g. cards), so that teams aren't wasting time making and re-making those bricks over and over again often in an inconsistent manner. It also contains some instruction manuals to help teams understand how to put pieces together, branding guides, etc. And now, all of these pieces can be used both by people and by AI to make digital interfaces.

Right now, as someone who would love to shift more into dedicated design system work, I'm betting on AI finally being enough for my company to set aside dedicated resources for even a small design system team. But it's very possible that the company is too stuck in their ways and still won't.

My current org does not have a design systems team yet as the product is very early stage

Strongly (it's a big advantage for us)

I'm currently in a company which is very early-stage and we still haven't set up a design system.

But I am currently having an active discussion with a design team on setting up the design system, what all to add and all of that stuff. Once the designs are ready, I think with AI what has changed is that migrating to the new design system would not be years of work, and we are planning to complete that within a month.

The design system is like the bricks in Lego. If you already have the bricks, creating the structure shown in the manual is really easy. Which is basically converting designs in Figma to actual production-ready code if we already have all the building blocks ready

If you need to measure the dimensions and create a brick every time you get a new structure, then you just have to spend a lot of time reworking. Which also leads to a loss of consistency.

I am not making any huge bet, but I feel that the teams who have leveraged the design system to like 90% perfectness would see an immediate productivity boost because of how AI references code and reuses it.

If you don't have a proper system in place, AI just tries to create new components or reuse existing ones by adding new dependencies and new conditional blocks, which leads to a mess and may lead to a lot of UX inconsistencies and bugs.

But if you have the building blocks figured out, all you just need to do is think about what to build rather than focusing on the actual implementation.

Leaning in hard

Strongly (it's a big advantage for us)

Hasn’t fundamentally changed as the DS owner, but there’s a huge focus in our vision and roadmap to align with the AI efforts of the rest of the company. The DS as an enabler for AI.

The core essence hasn’t changed: a collection of solved problems. The deliver mechanism may have changed, but the DS is still infrastructure.

If and when the AI bubble bursts and AI usage becomes infeasible due to pricing, our heavy focus on AI tooling may be an issue.

Experimenting carefully on the side

It seems across the industry that a lot of AI adoption pressure is organizational, which means I would imagine that the fate or existence of a design system or design system is oftentimes dependent upon its perception at the strategic/management level. So in that sense, I think that while “the design system is a necessary foundation for AI” can be a useful angle to protect the design system (and DS team), whether or not that plays out is far more about that perception management than it is about any sort of practical or tactical reality. I don’t think design systems are uniquely protected from the same assumptions/beliefs that cause management to cut roles for AI across any other function, regardless of whether it’s practically true that they perform a more foundational need.

I’ve been moving towards teaching people to fish, that is, focusing on platform agnostic guidelines, cross-platform alignment of terminology and methodologies, overarching guides to how and when to define semantic tokens, that sort of thing.

You’re responsible not for designing the product but building it at scale, and aligning that build across visual design tools and code tools.

That the right problems for our team to solve right now (on the Figma/documentation side) are in fact tool-agnostic, and about systems of naming, classification and communication. That
“tool-agnostic” also includes AI, and and organized system and well functioning team can take on AI just as any other consumer/stakeholder.

Experimenting carefully on the side

Not really (it's a comforting story)

Our last year was an exercise in rebuilding some trust lost during a recent initiative, where both parties involved (ourselves being one of them) mishandled how our design system should work. In the midst of this rebuild, AI exploded and the engineers moved on with their own direct Figma connections without UX or Design System included in that work. It's created a sort of panic initiative to try to get back ahead in tooling as a way to re-inject the "people who care" into the dev process which is clearly focused on output over quality.

A design system is documented design and development decisions, presented in multiple modalities, for consumption and generation of on-brand digital interfaces.

That we aren't in a total regression of the web, akin to the jQuery and Flash slide where the internet disregarded accessibility and performance for flashy.

Leaning in hard

Strongly (it's a big advantage for us)

We're leaning hard on AI and have spent a lot of time setting up the necessary "context."

The collection of tokens, rules, and patterns that create visual consistency across an organization's digital product or products.

My current bet is that speed won’t beat system quality. AI can generate an app fast, but I still think intentionally designed component libraries and patterns, particularly those grounded in accessibility and usability, will matter more long term.

I'm out of work, but keeping a close eye on what the community is saying/sharing. Previously, AI had no part to play in the DS I was working on.

Not really (it's a comforting story)

Just over 12 months ago, I had a job. Now I don't. ;-)

A design system is not a Figma library. Or a Storybook component library. Or a documentation site. It's all of these. And more.

Design systems need people who value craft & quality, not speed.

Leaning in hard

Somewhat (for now, but not for long)

As a product manager, I have to be thinking about what my users want from the design system. What they want, even more than ever, is speed, accuracy and ease of use. They know AI can help improve those things but are expecting that from the DS team. So, I'm evaluating our product at every turn and asking, "When and where can we and should we use AI as part of the solution?" It's A LOT of learning just to keep up with what's out there from a technical feasibility both for design and code.

I don't know that the "what" has changed for design systems. I think it's the "how" that has changed and continues to evolve.

I'm betting that we still need humans to own and manage the AI (context, data, results) of design systems. I may be wrong but gosh, I hope I'm right!

Skeptical, the hype outpaces the reality

Not really (it's a comforting story)

I was laid off in February after a very successful year and an excellent review. I have applied to more than 300 roles since then (both DS-specific and more broad end-to-end product design positions where contributing to the DS is only part of the duties), and have only gotten two interview offers. I'm being told by recruiters that getting between 500 and 800 applicants for an open role is the norm right now. I'm honestly unsure whether there is still a future in this field.

A design system is the structural layer that lets product teams ship consistent, accessible experiences without rebuilding the same decisions every time, and the operational discipline that keeps that layer healthy as the org grows.

My bet is that the current rush to integrate AI tools into design systems work is creating tech debt faster than it's saving time, and that by next year we'll start seeing the consequences clearly. Studies are already showing AI tools have minimal measurable impact on business productivity, but companies are using them to justify cutting skilled specialists. When the results don't show up in dollar signs, or when systems start breaking down, the people left on staff won't have the institutional knowledge or problem-solving skills to fix what's broken. Top-tier companies will have the resources to course-correct. I'm worried about how many mid-sized companies are going to find themselves in serious trouble. I could be wrong if the tools mature faster than I'm expecting, if the productivity gains start showing up in places we're not measuring yet, or if companies hold onto enough senior talent to weather the gap. But right now the math doesn't work for me.

Watching closely but not changing much

I hadn't thought about it that way

not a lot.
my org is investing in AI but we haven't fully implementing any MCP servers or meaningful connections to the code and models

so we use AI for documentation and day to day to task
but currently aren't building anything with agents

a collections of decisions aggregated and curated for consumption by both people and machines

that humans will always be needed

Watching closely but not changing much

Somewhat (for now, but not for long)

My actual role and responsibilities haven't changed much, other than having to increasingly field broad, vague questions like "Is the design system ready/structured for AI" from leaders who can't actually articulate what they mean or what their expectations of "ready for AI" entails.

I unfortunately don't have a re-definition yet. I sill define it the same way I always have.

That, in a large organization, one person or even just a few people, can't manage/govern it all. And, I'm pretty sure I'll be wrong about this in 1 year, giving what I'm already seeing some people (single person) do with AI and design systems.

Experimenting carefully on the side

Somewhat (for now, but not for long)

I'm trying to hold onto a process for creating and improving components based on product team needs, giving good defaults with hooks for customization. That is being challenged by AI using the system, which wants hard and fast rules where customization always open risk to mistakes. I can trust people to make good choices or reach out when they need help, but I can't trust AI to do tose things.

These pressures have me juggling both the process I know works for people with the question of what our role should be in the enablement of AI workflows. It feels like double-work, which is stressful on capacity. Also, the AI enablement work is the most uncertain whether the effort will pay off. We have to, due to organizational pressure, take bets that might end up being a total waste of time in 6 months.

A Design System is a practice of creating, maintaining and evolving tools that clarify for people and AI how we believe our products should look and function for our customers. These tools empower product teams and their agentic tools to apply those beliefs efficiently and consistently at scale across the product experience.

That our design system documentation will improve the quality of agentic output while reducing the token cost.

Further, we are betting people will still be using and (affording to use) these tools by then.

Leaning in hard

Somewhat (for now, but not for long)

My role has shifted intensely toward strategy work and determining how we will be AI ready and working differently by 2027.

Design systems roles are becoming less design and engineering and more strategy, project management, roadmapping and infrastructure mapping. We're at the center of figuring out how AI will design and develop consistently moving forward.

We're systems thinkers that are shifting to building interconnected databases and prompts for internal customers. I hope I'm wrong.

Experimenting carefully on the side

I hadn't thought about it that way

N/A

Man that is a GREAT question and I can't wait to hear everyone else's answers :P

That ours will be ok without leaning hard into AI yet.... :\

Leaning in hard

Somewhat (for now, but not for long)

It’s gone from having a somewhat clear vision of where we were going, and what excellence would look like, to navigating the moment and acknowledging that no one knows where we are going to end up.

Guiderails is a term that has been bouncing around in my mind lately. Like the bumpers along a the track for a splash ride. Everyone is floating freely in the same direction, propelled by this new technology, and the design system is there to bounce off of before going too far off course.

I feel pretty confident that structured data that describes the system (knowledge) and instructions that help perform specific, granular tasks (skills) will continue to be useful, but higher level concepts, full workflows, distribution mechanisms like marketplaces and MCP do not feel stable or settled. The models and tooling are still evolving fast and some of our higher level work is likely destined for the bin. 🤷

Leaning in hard

Somewhat (for now, but not for long)

My focus has wavered between documentation, Figma, and AI tools a lot, rather than than mostly on Figma where I was the most previously.

An interwoven system of systems that guides, instructs, and protects users, interfaces, and the sharing of branding, data, and content.

That they won't be entirely important in the shape they are now: UI-kits, documentations, codebases. AI will be able to infer much more and won't need Figma UI-kits as much, will share global sources of documentation, and will be able to build codebases so fast that "stale code" will be a thing of the past.

Leaning in hard

Strongly (it's a big advantage for us)

I mean... c'est moi. lol

I've gone from leading the design segment of design systems efforts on teams to being the designer and technical architect, and that shift has allowed me to understand and craft systems at a much deeper level.

The more I explore the technical space, the more I find myself able to create deeper layers of meaning, and deliver outcomes, to improve HX (Human Experience) and AX (Agent Experience).

A design system is an ecosystem of cultural and technical contracts between Humans and AI Agents to build complex systems together.

I mean... I don't really make bets, or invest effort, into areas where I think I'm wrong. So far, I've been right a lot even if I get a lot of choices wrong. lol

That said... I think I'm betting that native languages matter. The industry more and more wants to align itself to React, and I don't want to go down that road for a myriad of reasons. I've mitigated cross-platform development costs with smart architectural and technical choices that make separate multiple implementations trivial, but I'm certainly running the risk of being "purist over efficient".

Experimenting carefully on the side

Not really (it's a comforting story)

There are way more assumptions about not needing developers, designers, a design system, documentation, or as many people for any particular task due to misinformation and lack of personal experience or enough professional experience to understand quality and risk factors of the AI generated products.

A design system is a system of rules that form a consistent product or quality level based on a specific context need.

That the tech industry will come back after bad experiences and realize that they DO need design systems or some quality control at all or at least in order to be a good competitor.

Leaning in somewhat hard. Experimenting, but not as thoughtfully as we should. Its a bit of a free for all - everyone experiment, but no vision set, no coordination, just sort of chaos. We had gotten to a place where design and tech worked really well together in our process. With AI, we're regressing back to working in silos.

Not really (it's a comforting story)

As a people leader, not so much yet. Team has been experimenting, but we haven't yet instated a new way of work. But its coming soon. (We're a fintech, so often a bit behind the rest of the world.)

shrug

i'm struggling to see where we'll land. context. governance. ???.

sadly, that our team will continue to downsize. i do hope i'm wrong.

Leaning in hard

Somewhat (for now, but not for long)

I have moved to Ward research as a tool to refine our choices and even our expertise in our users. This releases pressure when there are conflicting opinions about how to move forward in a design decision.

A flexible system that creates build efficiency and visual consistency for a mobile experience.

That educating designers more about how to build a design system in ways that are close to CSS will pay off when we try to integrate AI

Experimenting carefully on the side

Somewhat (for now, but not for long)

Hasn't really changed.

A comprehensive framework to expedite and enforce/encourage consistent implementation that meets company goals across teams/products/etc.

Figma will be an important part of the system.

Leaning in hard

Not really (it's a comforting story)

Much, much less work in Figma, much, much more chasing prototypes that reference the design system, but don't build with it. Much, much more FOMO. Much more crunch, like we're all sprinting, but not sure towards what.

A shared methodology for how to design, build, and orchestrate consistent, quality experiences.

…hope I'm wrong about… That leadership won't see the value in having humans in the system loop, thinking that there are other mechanisms to obtain and propagate consistency and speed.

Leaning in hard

Not really (it's a comforting story)

I'm a one-person operation, so this is probably not the type of answer you're looking for. That said, everything I build now in my design system optimizes for AI agents, not human engineers.

I'm not sure I believe this, but I would say: A design system is code that represents predetermined design decisions, enabling AI agents to generate designs based on your companies design rules and strategies.

Brand designers, the people making creative decisions that form your brand-design POV, will become more valuable, as a required design system input.
The creative decisions -- typefaces, colors, spacings, transitions, etc -- that result in a distinctive brand, cannot be reliably authored by AI agents and typically are not assigned to design system teams.
I'm betting that:
- Brand designers will generate descriptive brand profiles to be consumed by AI agents.
- Design system teams will represent those descriptors in code code (.md files?)
- Product designers will use AI agents to author designs based on the design system and brand profiles.

Leaning in hard

Strongly (it's a big advantage for us)

As a designer the arrival of AI brought me back to work closer with code and reminded me of my early career where I was actually building the software I was designing. Having AI helping me with that has been tremendously empowering (and fun!) and made me learn a ton of stuff on how design systems are built in code.

What Skills, MCPs and agent are enabling is less chores, better overview, higher quality and giving everyone superpowers to do the things we never had time or resources to do before (update docs, create those components in the backlog we never had the time to build, review and clean our Figma files, improve accessibility etc.)

A design system is the shared foundational language companies use to build interfaces and is the context agents use when tasked to generate and build software or designs.

I am betting companies will use design systems as context for their products but what might happen is that companies will just generate "good enough" software that will be thrown out of the window 6 months later via another prompt.

I work as a Digital Designer adapting UX and UI practices including components and some design system foundations to e-commerce

Strongly (it's a big advantage for us)

From barely understanding it to thoroughly wanting to apply and learn more about it. It has become essential.

An series of instructions to build a digital product in mantainable and consistent way.

That AI will completely replace designers once the Design System is "done"

Leaning in hard

Somewhat (for now, but not for long)

Trying to wrangle non-designers and non-engineers vibe coding and vibe designing.

A standard toolkit for humans and AI to build consistent digital products.

Creating a custom allowed component utility component for Slots. Figma might release this as an official feature.

Experimenting carefully on the side

Somewhat (for now, but not for long)

Layered. I've always felt like there was a lot of "keeping all the plates spinning" between supporting teams using the system, requests for new components or patterns, identifying new patterns that need to be researched, documented, and implemented, and documenting patterns that already exist.

There's now just an additional layer of all that with having to consider how could AI use this information AND (from the leadership pushes) how could we use AI to do all of this faster. While also considering what "doing this faster" does to the human connection.

Get in, we're focusing on contextual experience.

Users are going to get sick of everything looking the same and will want advanced personalization options for how they define their product experience. Good thing foundational system work will enable this at scale.

Experimenting carefully on the side

Somewhat (for now, but not for long)

I teach in Design and Design system and I have had to quickly experiment with AI tools to articulate what new designers should be experimenting with both in Figma (ie variables definition) to claud code. Pivoting my Figma DS skills into MCP AI translation and prompting strategies.

Modern design systems are a massive set of interactive LEGO pieces that help everyone assemble things faster.

Design system roles will be expanded for the next few quarters, then undergo a large layoff cycle as systems become maintainable via AI.

Leaning in hard

Somewhat (for now, but not for long)

I'm spending more time writing component documentation for human usage and AI context. To be honest, I enjoy it just as much as designing. It's a good practice for us to maintain, because we're no longer kicking that can down the line, nor should we rely solely on AI to do so. Reasoning, documentation, and context should be unique to your design system, not someone else's.

The design patterns and infrastructure needed for any team to build cohesive, high-quality digital products.

Systems teams are the arbiters of efficiency and craftsmanship. That's easier to believe when leadership is on the same page.

Watching closely but not changing much

Strongly (it's a big advantage for us)

We've historically been used to strategizing, planning, and implementing all component creation and design system structure updates manually.

A bunch of branded component libraries that connect designers designs and developers code to the same source of truth.

That AI still needs design systems to provide accurate branded output.

Self-employed

Somewhat (for now, but not for long)

It seems like the ROI of design systems has become a little more visible thanks to the chaos of AI slop showing why it is important for companies to invest in DS to be able to ship features fast with AI.

Agentic-first has replaced consumer-first.

There will be an increase in the need for Design System AI Architects

Leaning in hard

Somewhat (for now, but not for long)

My DS team is on a sub system built to bridge the big corporate core DS to a set of products. We make patterns and templates that are closer to real product needs than the core is willing to go. The past few months AI has exploded in the design org. That has shifted us into building tools to simplify the process of creating vibe coded prototypes and demo apps that designers can use to iterate with quickly. One specific example is using our documentation to feed a DS and Patterns knowledge base that AI agents can read and interpret.

A design system is an evolving repo of product design knowledge to empower people (and bots) to build cohesive experiences.

I hope AI amplifies the need for design. If we can build anything quickly it should enable us to put a higher value and more time on early design and discovery phases. Bring back sketches and wireframes!

Experimenting carefully on the side

Somewhat (for now, but not for long)

I've really been pushing my team to explore how the design system team can get out in front of front-end and accessibility-related AI enablement for our designers and developers. We work in a highly regulated industry, so we've only just got access to Figma's AI tooling in the last month, and only have had access to Copilot in VS Code since the beginning of March, so we're doing a lot of catching up to the industry. One designer has been exploring Figma's AI tooling and documenting what juice is worth the squeeze and presenting on that to our adopting designers. The UX Engineers have been creating and modifying front-end and accessibility-related instructions, skills, prompts, and agent files that we make available to our adopting "full-stack" developers to guide their front-end development and the design system's coded components. TL;DR - My role now is making it as easy as possible for our adopters to use AI efficiently when it comes to adopting the design system.

Probably not much of a departure from a few years ago... A set of designed and coded UI patterns or components structured with design tokens and made available to design and development teams; supported by documentation on how and when to use each component along with content on accessibility, content design, and UI pattern examples.

That context must be provided via MCP servers, skills files, and explicit prompts.

Leaning in hard

Strongly (it's a big advantage for us)

We have gone from a list of loosely used variables to getting a team together to start our design system. AI has been a huge factor of getting leadership on board with having a design system

Bringing together the branding and culture of your system to share with your team in a standardised setting.

Hate to say but free expression in design is going to suffer, everyone will want to have a standardised design in their websites and applications and "something different" will become a luxury

Leaning in hard

Strongly (it's a big advantage for us)

Honestly, it's been a fight. Components now can be churned out, without thought or consideration, in an instant. These components are branded as _good enough_ because aesthetically they look great.

But we are fighting a battle whereby we have to justify the time and expense of what actually makes a design system and thats: A set of carefully considered settled solutions to problems in the UI/UX space.

The code is just the output of that but oftentimes now it's seen as the final product. That can't be further from the truth.

A set of carefully considered settled solutions to problems in the UI/UX space.

We are going to have a hard time justifying our existence. This is going to be the fight of our lives to prove that quality is paramount to quantity.

Skeptical, the hype outpaces the reality

Not really (it's a comforting story)

Change? I don't do change ;)

Filling the gap between brand teams and digital reality.

I say this, because many people have at least a vague understanding of what a brand team is in a large enterprise, so it gives me a starting point to try to explain what we do.

It's obviously an oversimplification, but I think of it in terms of "what would happen if we didn't have a design system" and _one_ of the (many) answers is every team would have to take brand's guidelines (aimed at glossy mags, billboards and packaging) and turn it into something that can be used in software. Everything else we do on top of that, is to serve that purpose.

We (well, not me) are drastically changing the way we create software, but not really changing the _kind_ of software we're creating. If we, by some miracle, find a way to make AI truly economical, environmentally friendly and ethical, then it's going to radically change the way we interact with technology. So, I'm gambling on that _not_ happening and even if we're using AI to create software, it'll still be the same kind of software we're creating today.

Experimenting carefully on the side

Somewhat (for now, but not for long)

Needing to support teams by providing DS context for AI tools.
Pressure to drop Figma and go directly to code / generate components using AI

A design system is a collection of reusable components, guidelines, and standards that teams use to build products (or experiences) consistently.

We will soon primarily be creating design systems for branded AI experiences vs for build product experiences. (The tools will be our primary consumers vs developers)

Experimenting carefully on the side

Somewhat (for now, but not for long)

This idea of context for AI vs documentation for humans is a good way to look at the change - I have to be on top of MANY new and constantly changing context scenarios.

The Design System covers the product design requirements, defines a contract for product managers to follow, and builds in the guardrails for engineering.

My company wants to change our design system drastically so it is hard to be sure. In general I'd say it will be dev platform agnostic, it can't all be about React and Tailwind.

Experimenting carefully on the side

Somewhat (for now, but not for long)

There's definitely been a rapid shift to embracing AI. The biggest thing I've recently discovered is that there are multiple independent efforts to try and get AI to build production-quality code with the design system but no one knows about each other's implementations. And as the person who built the system, I was frustrated/annoyed to find out this was happening without anyone asking me for my input. As design system practitioners, we're in the business of saying NO. But AI is in the business of saying YES, and also, "you're so smart." Hard to compete with that.

I describe it as a common language between design, engineering and product that describes how we make products at our company. More importantly, it's an agreement. We agree that this is what we call things, this is how we implement things, this is how we determine what goes in the system. Everything else, the UI kit, the design tokens, the component library... these are all just expressions of the system.

It seems like everyone is trying to figure out how to get deterministic outcomes with AI and design systems. I've been trying to "teach" Claude my design system methodology with some degree of success. I'm working on going freelance soon, which could either be disastrous or wildly successful, depending on how well I'm able to get my design systems to work with AI.

Skeptical, the hype outpaces the reality

Not really (it's a comforting story)

The story has often been about doing more with less, which isn't new. But what's challenging at the moment is it used to be "easier" to say "we'll do more with less...to benefit XYZ" but no one has clearly been able to articulate the real, defined, tactical, non-buzz-word benefit this time. These tools and erratic "we must shift to the market!" maneuvers are causing mountains of debt build up and consistently hurting the users we claim to care so much about.

My role hasn't changed--I am still the bridge builder, the designer-who-gets-tech, the 10k-foot view person, etc etc--but the audiences I've been expected to support have. And those audiences themselves (the ai tools/agents/etc) aren't well defined.

Come back to me on this one lol

Sticking to our foundations and not jumping at every pan-flash-new-technology is still the correct move for SYSTEMS that affect MANY.

Over the last couple years, we've approached AI pragmatically. Diving in evaluating tools and workflows across design and development teams to determine what was genuinely useful for our org’s maturity level.

I use tools like ChatGPT and Claude for system audits, component documentation, voice and tone guidelines, adoption reporting ideas, and evaluating newer design/dev workflows and tools like Figma Make, Magic Patterns, and Claude Design. Our UX engineer leveraged Claude Code heavily within DS monorepo and React component library workflows.

My overall view is that AI is becoming another important layer in design, but not a replacement for the collaboration, trust, and alignment that design systems still depend on.

Apart from the DS itself, I also used AI to assist other departments, helping product and marketing teams better understand tools like Userpilot, LogRocket, and FullStory by creating additional demo presentations, documentation, and workflow guidelines tailored to their needs that they could reference independently from the standard help docs from the tools.

I don’t think design systems “protect” anyone’s role outright, especially in a market where organizational and shareholder priorities can change quickly regardless of contribution or expertise.

I do believe that design systems become even more valuable in AI-enabled workflows because they provide context, guardrails, standards, and shared language for whatever is being generated, whether that output comes from a designer, developer, or AI tool.

With that said, I don’t view design systems as AI’s “necessary foundation,” or AI as the necessary foundation for a successful design system. The tools will continue to evolve either way.

What still matters most are the people making decisions around the system like defining standards, managing trade-offs, and determining what quality and consistency actually mean for the org. AI can accelerate production, but it still lacks the context and alignment needed to navigate the messy realities of enterprise systems work.

In my experience, that layer of decision-making becomes more important as the teams consolidate and tools become more capable.

After establishing my former company’s initial design system (2017) and later hiring its first UX Engineering positions (2022), my role became more focused on end-to-end process, communication, and adoption across business lines that still operated in pre-acquisition silos.

In the last 12 months, specifically, prior to my recent layoff in March, the day-to-day work itself didn’t change dramatically. A portion of my role was still listening to the folks adopting the system and those resistant to it, making trade-offs around patterns and governance, and continuing to champion the system through collaboration instead of enforcement.

One thing that hasn’t changed, and I mention in the other question, is that design systems are still mostly about people. The hardest part (especially in M&A scenarios) is creating alignment and building trust across teams toward a common goal. Some orgs embrace that quickly, and some struggle depending on leadership structure, team culture, and how design is valued within the business.

Even with all the changes happening right now, I believe design systems are becoming more important, not less.

Not much to redefine at its core.

"Yesterday, today, and tomorrow, a design system remains a scalable framework that creates consistency, reduces redundancy, and accelerates how teams build things together."

One bet I’m making is that, depending on the design maturity of the organization, design systems will continue moving through a few different stages. First they’re treated like side-of-desk passion projects, then officially funded initiatives, then internal products with dedicated teams and roadmaps.

But I think the next stage after that is even more interesting: design systems becoming less of a specialized design thing and more like standard operational infrastructure, treated similarly to any other core tool or platform within the company’s tech stack.

I think AI can and will help accelerate that mindset among leadership teams, but we’ll see.

Experimenting carefully on the side

Somewhat (for now, but not for long)

Reminding people why we have a system, managing expectations that we can simply 'turn on AI' tooling without major labour and experimentation.

The consistency layers which powers your product's infrastructure.

Designer system are necessary to keep AI consistent and on brand.

Skeptical, the hype outpaces the reality

Not really (it's a comforting story)

Mostly, it revolves around the maturing of the System itself. Thus providing support and insight. Also, as front-end is disappearing (no joke) in the organization, our community channels are becoming the go-to for all front-end and web development-related issues.

Its about the System and less about libraries and design kits. These are artefacts, but the strategic benefit of design systems remains.

Less about implementation and more about Knowledge consultation and support. Less investment in design and design implementation will have impact.

Experimenting carefully on the side

Somewhat (for now, but not for long)

I've primarily been a consumer, so it hasn't changed much outside of exploring how to leverage AI when using a design system in my work.

A source of truth that should align a team in approach, language, and structure.

AI will redefine (for the better) how we document our design systems and consume that documentation.

Experimenting carefully on the side

Somewhat (for now, but not for long),Strongly (it's a big advantage for us)

My company has strict security policies, so we’re much slower to adopt tools like Claude and Cursor. We’re just now exploring how we can leverage AI (GitHub Copilot) in our design systems work.

Design systems are collections of reusable patterns and assets used for building digital products.

Design systems will become even more formalized and generic, and will be available to install via a combination of AI and package managers like NPM.

Experimenting carefully on the side

Strongly (it's a big advantage for us)

With AI have a substitute of a companion wotking on the DS with me. I feel the pace increases and the expectations to use AI grow. I’m able to build components faster with AI assistancec

A design system is codified brand and product knowledge and decisions, providing context and building elements understandable to humans and AI.

I am unable to make any bet like that, as all might be untrue this time next year.

Leaning in hard

Strongly (it's a big advantage for us)

My teams remit has expanded and now I am responsible for AI systems and governance as well

They are the context engine and foundation of AI UI infrastructure

Product primitives

Experimenting carefully on the side

Strongly (it's a big advantage for us)

It changes very often as I work as a consultant in a consulting company.
So now with my DS day to day work, I do a lot of advisement and trainings

An organized scheme and method to give structure to all every conception (not just digital one) inside a company that support the whole brand experience at the end

That we need humans to craft a well thought design system talking architecture and tokens architecture as well

Experimenting carefully on the side

Not really (it's a comforting story)

We are using AI more to do the grunt work.

Collaborative with AI.

It should become an automated infrastructure.

Experimenting carefully on the side

Somewhat (for now, but not for long)

AI has obviously been a big part of the change, a lot of experimenting, trying to figure out how it fits into our workflow, and what the limitations are at our org. While it is more of a side effect of AI, it has felt like we are going to hang onto a lot of the design technologist responsibilities as well as the ops responsibilities whether AI sticks around or not. It feels like we are being asked to diversify skills at a rapid pace.

a design system is a working contract between eng and design that turns decisions into repeatable maintainable products

I think they are going to lay everyone off and regret it in 12 months when AI proves to be too expensive/doesn't replace designers.

Experimenting carefully on the side

Somewhat (for now, but not for long)

I've gone from focusing my attention on how humans consume our system (and fill in the gaps) to thinking about how to provide context for machines to read (and tightening to preemptively fill the gaps).

On one hand i feel unlocked. I am learning a lot and i have access to code and tools that I as a designer haven't had before (maybe just because i felt too nestled in my discipline). On the other hand i feel the space race (is it an arms race?) of these tools — it feels like its going to be an interesting but messy year at my org.

Design systems are infrastructure. Yes, AI is changing workflows, yes tools for production are evolving and changing, but at the end of the day the same thing is true today that was true yesterday — design systems are product infrastructure for teams building downstream. Its always been about quality and standards, those needs require dialing a little tighter.

It will take longer to get your organization AI ready than your design system and we'll be feeling the growing pains for a while to come.

Experimenting carefully on the side

Somewhat (for now, but not for long)

PMs have been using AI heavily to generate prototypes that they want the Design team to turn around and incorporate. Us DS Engineers are looking to help sort out this disrupt in workflow by establishing AI rule/guidelines and skills to streamline the ingestion for the Design team.

Brad Frost has already stated that "Design System" should instead be referred to as "UI System". I have no idea how to frame what's happening today. It's a brave new world in Design Systems.

That the Design team should remain the expert reviewer/approver for all the new prototypes that are coming their way these days. In one year, this step could be reduced to robust guardrail that consists of rule/guidelines written and maintained by the DS Team.

Leaning in hard

Strongly (it's a big advantage for us),Once we finish filling all the gaps in that foundational layer, I think our ability to think at the system level will be transferrable to the future state of design systems work. The work may shift from only working on the system, to working on the tooling and enablement side of things.

I’ve seen a strong shift towards documentation in the past 12 months. I’m either writing it directly or working on AI tooling to help me analyze and write it. Even when I’m working on other things like Figma Make templates and kits, much of the work has to do with connecting to some form of documentation.

The agreed context decided by an organization for how to create consistent and high quality experience for end users.

Systems teams will be increasingly expected to build tooling to enable more appropriate use of the system by humans and AI agents.

Leaning in hard

Strongly (it's a big advantage for us)

I think I am wearing many more hats. I think roles are converging, the divide between engineering, design and product is getting smaller.

A design system is the structured source of truth for product decisions, so design stays executable by both humans and AI at scale.

I think the source of truth for design systems is moving more towards code and out of tools like Figma.

Leaning in hard

Somewhat (for now, but not for long)

Coming into 2026 I thought that our design system team would (a) need to embrace using AI in our work (something that is being pushed across the company) and (b) we would need to establish guidelines / patterns for designing AI experiences. In the last 5 months, i've realized that I was missing a much bigger part (c) - that as the design system team we need to equip the design system to be fed into all the AI tooling our company is experimenting with

The source of truth for how our company designs and develops user experiences

That our investment in AI tool integration will pay off and won't be wasted effort when the tools inevitably change

N/A

100