I recently shared my opinion on using #AI: buff.ly/wltTmXg Here are more use cases buff.ly/UNyNpZw like: - brainstorming & concept exploration - interactive prototypes for early testing But: - Require refinement - introduce inconsistencies

Great #tool to generate #accessible #colorScales buff.ly/WnyH49r I built something similar at work, but never got around polishing it enough to open source it. Great to see @seanw.org did. ❤️

I'm definitely a bit #AI skeptical but some things, especially assistant AIs are great already. This is a great perspective on this: buff.ly/Y0nlPtz We are very early in the AI game. There is a lot of overhype at the moment, a lot of fear, but AI will definitely have a big impact in the long run.

Never test multiple designs? buff.ly/0olDGTy by @adamsilverhq.bsky.social Testing multiple versions: - unclear results with no winner - needs more resources - users may not care about the difference Instead: Design & test one version to uncover its issues, iterate and improve until it is great.

Interesting: According to @rosswintle.uk software CAN be finished: buff.ly/htPAEhE For example embedded software, like on a home appliance or classic video games like N64. “Finished” used to mean it has all planned features, is secure, and works without updates. Is this still possible today?

In buff.ly/cMUML9y, @hvpandya.bsky.social argues that speed is not the same as haste. Speed is high velocity without sacrificing quality. If you can achieve this, it is a game changer. Being fast allows to shape the agenda. Realize what matters and act on it.

My #styleDictionary utils package now fully supports #w3c #designTokens allowing you to easily transform your w3c tokens with styleDictionary: Check it out: buff.ly/554KaMM And let me know how it is working for you.

Interesting 💭 buff.ly/xkkOVw5 by @joemag.bsky.social #AI reshapes tools & mental models involved in coding, nudging the industry toward valuing clarity of requirements, high-quality tests, and careful review over manual code authorship. True or do we shoot us in our own foot?

#Documentation is a #shared responsibility buff.ly/N4AyfLk Involving all disciplines into #DesignSystem #docs avoids misalignment and creates buy-in for docs.

Reminder buff.ly/eN5Lw0r to start small & don't copy #FAANG from the start. A monolith works well for small teams, don't start with microservices. Consider the costs of your decisions like reduced output due to complexity. Build for YOUR needs and size.

Love buff.ly/FsKeJlD Some important thoughts: - risk: irrelevance > tech debt - good enough ≠ sloppy - tech debt = trade-off for faster learning - Never ship code that is insecure, could lose user data, is broken - Shipping fast → options: pivot on real feedback - Refactor once you hit market-fit

We have an ambivalent relationship to #friction. Sometimes we reduce it, than we argue for its merits again. @phirephoenix.com argues for it in her post: buff.ly/ImPkFu9 The final argument: friction, between people and ideas, is what sustains democracy, creativity & empathy.

Insightful post buff.ly/P4idzdZ on how #AI changes the way software is built. With #AI agents engineers get working features out a lot faster. Often before we can define the #UX. But instead of complaining, embraced this. We can play with a working feature to explore how to optimize the UX for it.

#LiquidGlass is problematic, according to @nngroupux.bsky.social buff.ly/b2Fy7fH I have noticed some of those and other issues on mac as well. I am very surprised by the decline in quality at #apple. They used to be the #UX pinnacle and now they seem to overfocus on style. Any idea why?

Some time ago I wrote about building software for personal use: buff.ly/s1QfE6H Karan Sharma now approaches this idea in combination with #AI buff.ly/t8Mg9K3 which opens it up to a larger audience without prior coding knowledge. It's a great way to get into AI too.

An alternative to #StyleDictionary: terrazzo.app Haven't tried it. But it sounds interesting, especially that it is built with w3c tokens as the one and only input. I guess it all is up to how easy it is to build plugins and if you can easily get to all parts in the built process. WDYT?

The long awaited first stable version of the #designTokens spec has been published recently: buff.ly/Utcx1jl This is a very important step for the community as it means tools can now work against a standard. I expect a lot of the existing tools to be updated soon.

Finally @joshwcomeau.com shows us how to get native spring and bounce #animations in #css: buff.ly/vncwbGI The performance impact is neglectable in most cases, and the result is fantastic. Using css variables reduces file size and makes it easier to reuse your spring animations

Exciting! @figma.com release extended collections, finally multi-branding is easy in Figma. buff.ly/6wdXnPN Last year I did a whole day #designToken workshop @hatch and multi-brand was a big part and such a pain. Happy to see it solved (hopefully). Going to explore it now.

Its interesting to see how companies work on their icon systems: buff.ly/ojaHdyh Especially product icons, which need to without a label, e.g. in the dock. They must be recognizable, feel like being part of a set but still different enough so users understand which specific product it is.

Filter: simple until you build them buff.ly/B0GOBuu Great filters empower users, reduce frustration, and boost the usability. Bad filters lead to frustration & churn Great filters: - prevent errors - allow sharing a filtered states - are easy to find and use

Interesting and rarely discussed: designing for #ADHS buff.ly/GFmwbDn ADHD often includes forgetfulness, impulsivity, distractibility, hyperactivity, time-blindness, and sensitivity to sensory overload. Empathetic, flexible design can mitigates ADHD-related challenges.

Some non-typical #designPrinciples buff.ly/1WuvsKl - Ensure Easy Access: Clear, predictable home link & simple URLs - Communicate Who You Are: prominent logo - Reveal Content Through Examples: important offerings above the fold, - Prompt Actions & Navigation: clear, descriptive, & prominent CTAs

I stumbled upon this buff.ly/Qm3bbgZ by @fabriciot.bsky.social again and it resonated a lot. Principles for building better products: - Less, but better - Start small - Use data - Feature budget: Limit total # of features - Say no: Decline non-essential features - regularly reassess your product

#AI & #vibeCoding is often not perfect. Shipping AI work to production is risky. But that is not the same for personal tools. In my article buff.ly/wr3Ppnn I describe how you can explore coding with AI by creating tools that make your live easier by doing mundane, repetitive tasks for you.

Good tip @donnie.damato.design: buff.ly/ysu28eJ Don't use `onDark` in your tokens, this requires adding another set of tokens for everything on dark. Better define a dark theme and use this in the section that needs a dark BG. This way no additional tokens are needed and it works out of the box.

I started working on an #AI project, so this comes just in time by @smashingmagazine.com buff.ly/qSqSzSL AI must - show ability to accurately perform tasks - gain trust by operating transparently - generate predictable output You should communicate limitations of AI to users.

Just re-read @mattstromawn.bsky.social: buff.ly/xqE8kN9 - #DesignSystems are moving from static “dictionaries” to dynamic “functions” - Dynamic logic enables adaptive, personalized, and accessible UIs - Tools like CSS clamp() and Material’s dynamic color make this future possible

Worth a re-read: buff.ly/cHyYAbs > A thriving DS requires genuine passion from its builders & users What makes a #DesignSystem Loved? - exceed expectations - provide reliability - allow customization to adapt to user needs - buy-in through steering committee - get insights from user Champions

Great post on how to define your target #audience without just basic demographics: buff.ly/s3xEVQ2 by @asmartbear.com Much more important aspects are: - common problems - buying constraints Which you can get from customer interviews. Another overlooked aspect is motivations

Interesting insight into #slack's #designOps team buff.ly/ZvDiGpx They drive impact by focusing ruthlessly on company goals and making sure their programs scale beyond individual teams. I assume they have a good connection to leadership to pull this off and can make some tough calls on what to do.

Design to avoid serial task switching buff.ly/w1VTlHn Serial task switching significantly reduces efficiency, quality & cognitive well-being. When designing products, reduce unnecessary task switching & rather promote focus and sustained attention. This yields better outcomes for your users.

Great point buff.ly/Pkq8rcR by @martinfowler.com on why expert generalists are an important part of any team. Are you one? Key skills of expert generalists are: - Curiosity - Collaborativeness - Customer Focus - Fundamentals-Driven - Sympathy for Adjacent Domains - Blend of Depth and Breadth

How can a #designSystem up-level your companies design quality? Find out in my latest article: buff.ly/KFlL5C9 I discuss how to define goals that lead to higher design quality and how to evaluate your design system against those goals.

In buff.ly/tMSgiG7 @alexanderkondov.bsky.social states that pushing vibe code (in contrast to AI assisted code) is problematic because: - it ignores existing code & team conventions - makes understanding the code harder (maintenance) - you did not consider possible current or future side-effects

Interesting insights into @HP 's effort to unify their icon system: buff.ly/P55VhlF If this is something you are currently working on, there are definitely a few pointers you can take from their experience: - audit & designer interviews - lessons from the audit - Icon style: unified but flexible

Deviation from the #designSystem is not failure, it is a signal by @bencallahan.com buff.ly/kcgfTrr Takeaways: - Many deviations in the same direction point to a gap in your system - Use this to engage with users, listen & find the what & why - partner up to resolve the gaps

While #AI can increase productivity, it may have negative long term effects on an individual level: buff.ly/hDxqj2w This MIT study shows that relying on LLMs for essay writing weaken brain connectivity, cognitive engagement, and long-term learning outcomes. → Use LLMs with caution

What impact does #AI have on #design and #language? While AI makes it easier, especially for folks new to a field and overall increases productivity, the quality often declines. Design and content become more generic, suffering from sameness and uncreative solutions. →

The #EAA makes #accessiblility a legal requirement for digital products and services. Your #DesignSystem plays a major role in assuring your products can be accessible by June 2025. Find out about the requirements and how to address them with you DS:

Visual #DesignSystem usage #audit: buff.ly/LnRiaS4 @tonyward.dev Screenshots show the page & context a component is used. Highlighting components tells a strong story: All red components are from the design system. It's easy to see adoption gaps, inconsistencies & overrides.

#DesignSystem #Component conformance with #wcag, is it possible? → buff.ly/BwaamLl Conformance ensures that every component instance upholds the design system’s standards for accessibility, quality, and consistency—at scale and over time.

A fantastic @figma.com #plugin for #designSystems buff.ly/NaZrtDy You can scan one or all pages in your DS library to find missing doc links, descriptions & even better, values without assigned variables (#designTokens). In just 5 min I fix a bunch of missing tokens. 🔥

#AI is raising the floor, not ceiling according to buff.ly/k01z1fc AI does not automate mastery, but democratizes basic competence. In the long-term, shortcuts don’t produce real mastery. It's great to get up to speed with something, but from there it is up to you.

A very valuable perspective on #vibeCoding with #AI: ilo.im/165uaf by @val.town In Short: Vibe coding is creating bad code that no one understands. It is fine for prototypes (legacy code is only code that is maintained), but it is a very bad idea for real production code.

Grug believe in "make → test → cry → fix", the ultimate design cycle: buff.ly/EoyhLS4 A fun little read about basics of good design. Also, a funny bot to chat about your design ideas with.

How do #AI #agents create a #ROI for you? Find out: buff.ly/46oRtqU TL;DR - Labor efficiency: Cost savings and automation. - Net-new revenue: Generating value from untapped opportunities. - Optimization: Continuous, holistic business transformation.

Learn how to create an #accessible #svg buff.ly/uoTEISr by @a11y-collective.com Common Pitfalls: - Missing / vague descriptions for informative graphics - No ARIA roles or incorrect attribute use - Relying on color as the only means of communication - Interactive elements not keyboard accessible

#AI thoughts buff.ly/rsGGwli AI has shifted us from "design to build" to "prototype to productize". Designers are now slower than eng and come into a live product to improve ux. On the positive this means we can build better product through rapid iteration of the actual product, with real data.

#Content for fun vs. purpose must be created differently. - Fun: Surprising, delightful, novel or playful interactions - Purpose: Expected, clear, fast & straightforward interactions Both content types could be in the same product, depending on the user intend. buff.ly/x5NQuyq by @automattic.com