Lukas Oppermann
@lukasoppermann.com
Design @Github / opinions are my own
#DesignSystems #DesignTokens #UX #accessibility
💭 writing http://medium.com/@lukasoppermann
☕️ coffee lover
🐥 dad
(he/him)
#DesignSystems #DesignTokens #UX #accessibility
💭 writing http://medium.com/@lukasoppermann
☕️ coffee lover
🐥 dad
(he/him)
#DesignTokens finally got a spec, here is what changed buff.ly/YyLttp2 by @zeroheight.com
- stable json format
- mode / theming support
- modern color formats (like oklch & display p3)
- separate alpha value for tokens
- referencing of tokens
Sadly still no way to change alpha for a ref
- stable json format
- mode / theming support
- modern color formats (like oklch & display p3)
- separate alpha value for tokens
- referencing of tokens
Sadly still no way to change alpha for a ref
What’s new in the Design Tokens spec: From static to living design data - zeroheight
The Design Tokens spec reaches its first stable version, introducing modern color spaces, groups/aliases, and token resolvers to make design systems scalable and cross-platform.
buff.ly
February 9, 2026 at 5:33 PM
#DesignTokens finally got a spec, here is what changed buff.ly/YyLttp2 by @zeroheight.com
- stable json format
- mode / theming support
- modern color formats (like oklch & display p3)
- separate alpha value for tokens
- referencing of tokens
Sadly still no way to change alpha for a ref
- stable json format
- mode / theming support
- modern color formats (like oklch & display p3)
- separate alpha value for tokens
- referencing of tokens
Sadly still no way to change alpha for a ref
Interesting thoughts on the #future of #designSystem #components buff.ly/B9APeii
Components as data. The benefits: components can be generated per platform, AI can easily build them. The down side: files get very large and complex, as do the workflows that build components from the data.
Components as data. The benefits: components can be generated per platform, AI can easily build them. The down side: files get very large and complex, as do the workflows that build components from the data.
Components as Data
How to define components independent of platforms to scale a system’s impact
buff.ly
February 6, 2026 at 5:45 PM
Interesting thoughts on the #future of #designSystem #components buff.ly/B9APeii
Components as data. The benefits: components can be generated per platform, AI can easily build them. The down side: files get very large and complex, as do the workflows that build components from the data.
Components as data. The benefits: components can be generated per platform, AI can easily build them. The down side: files get very large and complex, as do the workflows that build components from the data.
This article buff.ly/NrmWMJe argues that code comments are key for AI to understand code usage.
- Focus on intent, constraints & “why,” don't restate the code!
- Business rules & domain context.
- Edge cases, failure modes & non-obvious performance constraints.
- Focus on intent, constraints & “why,” don't restate the code!
- Business rules & domain context.
- Edge cases, failure modes & non-obvious performance constraints.
The Importance of Code Comments for Modern AI Coding Assistants
In an era where AI coding assistants have become indispensable development tools, the role of code comments has evolved from a simple best…
buff.ly
February 4, 2026 at 5:26 PM
This article buff.ly/NrmWMJe argues that code comments are key for AI to understand code usage.
- Focus on intent, constraints & “why,” don't restate the code!
- Business rules & domain context.
- Edge cases, failure modes & non-obvious performance constraints.
- Focus on intent, constraints & “why,” don't restate the code!
- Business rules & domain context.
- Edge cases, failure modes & non-obvious performance constraints.
Static letter-spacing breaks responsiveness—what works in print can harm web readability. @cloudfour.com reveals how to adapt letter-spacing for fluid, #responsive typography:
Responsive Letter Spacing
Minimizing the readability impact of a typographic brand requirement.
buff.ly
February 2, 2026 at 5:33 PM
Static letter-spacing breaks responsiveness—what works in print can harm web readability. @cloudfour.com reveals how to adapt letter-spacing for fluid, #responsive typography:
Increase your #AI #productivity by running multiple local #agents on the same repo at the same time.
How?
Just read my post on using Git worktrees to make this possible:
→ buff.ly/xSwxvJp
Safely run agents on different branches without any risk of conflicts. Especially nice for long running tasks
How?
Just read my post on using Git worktrees to make this possible:
→ buff.ly/xSwxvJp
Safely run agents on different branches without any risk of conflicts. Especially nice for long running tasks
Running Multiple AI Agents at Once Using Git Worktrees
Or how to get more productive with long running agents
buff.ly
January 28, 2026 at 5:26 PM
Increase your #AI #productivity by running multiple local #agents on the same repo at the same time.
How?
Just read my post on using Git worktrees to make this possible:
→ buff.ly/xSwxvJp
Safely run agents on different branches without any risk of conflicts. Especially nice for long running tasks
How?
Just read my post on using Git worktrees to make this possible:
→ buff.ly/xSwxvJp
Safely run agents on different branches without any risk of conflicts. Especially nice for long running tasks
Use flow maps to visualize system problems. It's the best way to externalize your thoughts and ensure team alignment. buff.ly/NGlKo7W
Bringing your thoughts on paper uncovers different understandings that would have gone unnoticed if you just talk about it.
Bringing your thoughts on paper uncovers different understandings that would have gone unnoticed if you just talk about it.
Mapping what is in my head: turning invisible solutions into shared clarity
In my head it’s all crystal clear. On a whiteboard? A mess of Post-its. Mapping is how chaos becomes clarity.
buff.ly
January 26, 2026 at 5:33 PM
Use flow maps to visualize system problems. It's the best way to externalize your thoughts and ensure team alignment. buff.ly/NGlKo7W
Bringing your thoughts on paper uncovers different understandings that would have gone unnoticed if you just talk about it.
Bringing your thoughts on paper uncovers different understandings that would have gone unnoticed if you just talk about it.
I'm thinking about #Component composability atm, @zeroheight.com's
post buff.ly/BECCVrr came at the right time
Benefits:
- supports diverse products & edge cases
- Fewer one‑offs; teams can adapt components
- Feedback: real‑world compositions show which patterns to promote
post buff.ly/BECCVrr came at the right time
Benefits:
- supports diverse products & edge cases
- Fewer one‑offs; teams can adapt components
- Feedback: real‑world compositions show which patterns to promote
Using composability over inheritance to scale design systems - zeroheight
Adam Sedwick explains why you should choose composability over inheritance to scale design systems, turning them into a resilient toolkit teams actually use.
buff.ly
January 23, 2026 at 5:45 PM
I'm thinking about #Component composability atm, @zeroheight.com's
post buff.ly/BECCVrr came at the right time
Benefits:
- supports diverse products & edge cases
- Fewer one‑offs; teams can adapt components
- Feedback: real‑world compositions show which patterns to promote
post buff.ly/BECCVrr came at the right time
Benefits:
- supports diverse products & edge cases
- Fewer one‑offs; teams can adapt components
- Feedback: real‑world compositions show which patterns to promote
Design tokens finally are stable with W3C design token standard (2025.10). 💎
Learn why this matters and how to
✅ Automate migrations with AI
✅ Setup your own w3c #designToken pipeline
✅ Make it easy with style-dictionary-utils
✅ future-proof your system
→ Get started now:
Learn why this matters and how to
✅ Automate migrations with AI
✅ Setup your own w3c #designToken pipeline
✅ Make it easy with style-dictionary-utils
✅ future-proof your system
→ Get started now:
Design tokens with confidence
Why the W3C design token standard is your new foundation.
buff.ly
January 21, 2026 at 5:26 PM
Design tokens finally are stable with W3C design token standard (2025.10). 💎
Learn why this matters and how to
✅ Automate migrations with AI
✅ Setup your own w3c #designToken pipeline
✅ Make it easy with style-dictionary-utils
✅ future-proof your system
→ Get started now:
Learn why this matters and how to
✅ Automate migrations with AI
✅ Setup your own w3c #designToken pipeline
✅ Make it easy with style-dictionary-utils
✅ future-proof your system
→ Get started now:
In love with this description of good execution by buff.ly/cSM8gTH
Signs of Good Execution:
- Low operational noise
- Calm, manageable com channels
- Steady delivery & reliability metrics
- Teams correcting quickly, with visible progress
- Trust & clarity replacing chaos & crisis management
Signs of Good Execution:
- Low operational noise
- Calm, manageable com channels
- Steady delivery & reliability metrics
- Teams correcting quickly, with visible progress
- Trust & clarity replacing chaos & crisis management
What Good Execution Looks Like
Quiet execution means low operational noise, clear ownership and stable context, with delivery, reliability and human signals aligned.
buff.ly
January 19, 2026 at 5:33 PM
In love with this description of good execution by buff.ly/cSM8gTH
Signs of Good Execution:
- Low operational noise
- Calm, manageable com channels
- Steady delivery & reliability metrics
- Teams correcting quickly, with visible progress
- Trust & clarity replacing chaos & crisis management
Signs of Good Execution:
- Low operational noise
- Calm, manageable com channels
- Steady delivery & reliability metrics
- Teams correcting quickly, with visible progress
- Trust & clarity replacing chaos & crisis management
I recently wrote about #designQuality buff.ly/zUdXwCY so naturally #NielsonNormanGroup writing about the same idea is very interesting too me:
→ buff.ly/BkvUDCe
They focus on alignment, hierarchy, intentional colors & consistent, with interesting examples.
→ buff.ly/BkvUDCe
They focus on alignment, hierarchy, intentional colors & consistent, with interesting examples.
What is design quality?
A closer look at what designers should strive for — and why it’s important.
buff.ly
January 14, 2026 at 5:26 PM
I recently wrote about #designQuality buff.ly/zUdXwCY so naturally #NielsonNormanGroup writing about the same idea is very interesting too me:
→ buff.ly/BkvUDCe
They focus on alignment, hierarchy, intentional colors & consistent, with interesting examples.
→ buff.ly/BkvUDCe
They focus on alignment, hierarchy, intentional colors & consistent, with interesting examples.
AI coding agents should be managed, guided & reviewed like very fast but junior engineer according to @chrisloy.bsky.social buff.ly/Ssn94q3
- AI Changes the Coding Process: Less coding, but more fixing
- Observed gains are a lot smaller than promised, 10% rather than 10x
- AI Changes the Coding Process: Less coding, but more fixing
- Observed gains are a lot smaller than promised, 10% rather than 10x
The AI coding trap | Chris Loy
If you ever watch someone “coding”, you might see them spending far more time staring into space than typing on their keyboard.
buff.ly
January 12, 2026 at 5:33 PM
AI coding agents should be managed, guided & reviewed like very fast but junior engineer according to @chrisloy.bsky.social buff.ly/Ssn94q3
- AI Changes the Coding Process: Less coding, but more fixing
- Observed gains are a lot smaller than promised, 10% rather than 10x
- AI Changes the Coding Process: Less coding, but more fixing
- Observed gains are a lot smaller than promised, 10% rather than 10x
Mission, vision, North Star–it's often just jargon. To be useful, make them authentic & actionable. Clarity beats buzzwords: choose a purpose that’s real, a plausible vision with a timeframe, and milestones you can actually measure.
Mission, Vision, poTAYto, poTAHto
Mission, vision, purpose, BHAG. Are they useful north stars, or academic nonsense?
buff.ly
January 9, 2026 at 5:45 PM
Mission, vision, North Star–it's often just jargon. To be useful, make them authentic & actionable. Clarity beats buzzwords: choose a purpose that’s real, a plausible vision with a timeframe, and milestones you can actually measure.
#Motion systems that "fail safely" by @adobe.com
Defensive animation is key for motion-sensitive users: cap duration, avoid infinite loops, provide emergency stops, honor “prefers-reduced-motion,” & offer static fallbacks. Accessible, resilient motion is good design:
Defensive animation is key for motion-sensitive users: cap duration, avoid infinite loops, provide emergency stops, honor “prefers-reduced-motion,” & offer static fallbacks. Accessible, resilient motion is good design:
Animation that fails safely: Defensive design for motion-sensitive users
A practical framework for designing and engineering safe, resilient, and inclusive motion
buff.ly
January 7, 2026 at 5:26 PM
#Motion systems that "fail safely" by @adobe.com
Defensive animation is key for motion-sensitive users: cap duration, avoid infinite loops, provide emergency stops, honor “prefers-reduced-motion,” & offer static fallbacks. Accessible, resilient motion is good design:
Defensive animation is key for motion-sensitive users: cap duration, avoid infinite loops, provide emergency stops, honor “prefers-reduced-motion,” & offer static fallbacks. Accessible, resilient motion is good design:
New #FluidType by @matthiasott.com approach that deals with the issue of users changing their base font: buff.ly/OeXELrC
By introducing a damping factor, if the base font gets larger, the increase per step gets smaller avoiding overly large font sizes. Be aware of currently limited browser support
By introducing a damping factor, if the base font gets larger, the increase per step gets smaller avoiding overly large font sizes. Be aware of currently limited browser support
Compressed Fluid Typography · Matthias Ott
Matthias Ott is an independent user experience designer and web design engineer from Stuttgart, Germany. He also teaches Interface Prototyping at the Muthesius Academy of Fine Arts and Design.
buff.ly
January 5, 2026 at 5:33 PM
New #FluidType by @matthiasott.com approach that deals with the issue of users changing their base font: buff.ly/OeXELrC
By introducing a damping factor, if the base font gets larger, the increase per step gets smaller avoiding overly large font sizes. Be aware of currently limited browser support
By introducing a damping factor, if the base font gets larger, the increase per step gets smaller avoiding overly large font sizes. Be aware of currently limited browser support
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
Here are more use cases buff.ly/UNyNpZw like:
- brainstorming & concept exploration
- interactive prototypes for early testing
But:
- Require refinement
- introduce inconsistencies
Good from Afar, But Far from Good: AI Prototyping in Real Design Contexts
AI prototyping tools follow general directions but lack the judgment and nuance of an experienced designer.
buff.ly
January 2, 2026 at 5:45 PM
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
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. ❤️
InclusiveColors - Accessible color palette creator
Try it now to create your own color palettes!
buff.ly
December 31, 2025 at 5:26 PM
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.
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.
AI's Dial-Up Era
It is 1995.
buff.ly
December 29, 2025 at 5:33 PM
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.
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.
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.
Why I don’t test different designs at the same time
Adam Silver – interaction designer - London, UK
buff.ly
December 26, 2025 at 5:45 PM
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.
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?
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?
Software can be finished - Ross Wintle
You probably don't want to build "finished software". But what can we learn by thinking about it as a concept?
buff.ly
December 24, 2025 at 5:26 PM
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?
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.
Being fast allows to shape the agenda. Realize what matters and act on it.
buff.ly
December 22, 2025 at 5:33 PM
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.
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.
Check it out: buff.ly/554KaMM
And let me know how it is working for you.
GitHub - lukasoppermann/style-dictionary-utils: A package of transforms, formats, filters, etc. to use in your style dictionary config to transform your design tokens
A package of transforms, formats, filters, etc. to use in your style dictionary config to transform your design tokens - lukasoppermann/style-dictionary-utils
buff.ly
December 19, 2025 at 5:45 PM
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.
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?
#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?
buff.ly
December 17, 2025 at 5:26 PM
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?
#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.
Involving all disciplines into #DesignSystem #docs avoids misalignment and creates buy-in for docs.
Writing in Sync: Design System Content
Design systems help create a shared language within an organization, but without good documentation, no one will be able to speak that language.
buff.ly
December 15, 2025 at 5:33 PM
#Documentation is a #shared responsibility buff.ly/N4AyfLk
Involving all disciplines into #DesignSystem #docs avoids misalignment and creates buy-in for docs.
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.
Build for YOUR needs and size.
Why is everything so scalable? - Stavros' Stuff
I’m entirely convinced that basically every developer alive today heard the adage “dress for the job you want, not the job you have” and figured that, since they always wear jeans and a t-shirt…
buff.ly
December 12, 2025 at 5:45 PM
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.
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
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
Good Enough Is a Strategy
Your competitors will eat your lunch while you refactor
buff.ly
December 10, 2025 at 5:26 PM
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
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