Anthropic's new design tool fills the exploration gap in the Claude Code workflow. But if you think this means AI does the designing now, you're about to ship a lot of very expensive mediocrity.
DoW operations and AI integration is one of the least discussed but highest-stakes enterprise deployments happening right now. The standard procurement cycle is completely mismatched with the pace of capability change. How are you seeing acquisition frameworks adapt to evaluate AI systems that could be functionally different by the time the contract is signed?
Honestly, I think most of the debate is anchored on the wrong question. Everyone’s arguing which LLM to license when the answer is which garden to plant in.
AI Foundry, Bedrock, Vertex, and the orchestration layers being built on top of them are what actually solve the procurement-vs-capability mismatch. You stop procuring a model and start procuring optionality, governance, observability, guardrails, and the ability to swap the backbone when a better one arrives. Contracts specify outcomes and SLAs, not model names.
The model becomes a substitutable component. The durable value sits in the layer that composes and governs it: orchestration, not decoration.
Really needed this. The constant AI hype is exhausting, and I agree completely! So much of the “design is cooked” and “you can make everything in minutes” discourse creates a false illusion of how designers should work, when taste, judgment, and real design thinking still matter most.
Thanks for that more nuanced look at your perspective on Claude Design. What I find interesting is that you still see Figma as part of the design production process. But isn't Figma exactly the tool where interaction states, edge cases, and viewports are the most cumbersome to represent, and where inconsistencies between design and code are most likely to happen?
Why shouldn't I develop my design system directly in a live repository at the code level in the future, instead of turning static images into code and vice versa? Personally, as a designer, I don't see much point in continuing to work in static layout environments when I'm designing digital and interactive products... For quick exploration or iterating through a few module variants—it's great for that. But for design system management, I don't see a future for the tool here.
Second point: If everything looks the same, how do brands and products differentiate themselves? I think the human comparison works well here: If everyone were equally good-looking, who would be more successful in the end? The person with the most pleasant personality and story. And I believe that applies to products and brands too. Design excellence is, as you correctly wrote, now free. Everyone can look good. And the point will come where AI is able to produce results that are less sloppy and more individual. But everything will predictably be pretty polished. I believe the only differentiator for brands can be their personality and story, rather than their visual appearance—because even that (in terms of brand design) is becoming increasingly similar and interchangeable.
One more thing: when working alone whats your strategy for reiterating designs based on feedback? What kind of measurement tool and do you automate reports for it with some tresholds that will trigger a deeper investigation into a friction point ?
Interesting view, as a former art director/creative from the ad industry and a designer I see that these two branches is merging. What are your take on visuals, third party and native tools for creating content for webdesign? The combination of a designers intuition and the creatives ability to create different concepts in form of visualizations in form of animation, images and video will matter more than the layout. But that’s my two cents on the matter. Im not a seasoned designer so I’ve never followed a particular design system or anything like that.
I am so happy to have this perspective from someone who works in design. Thank you for all of this great insight 🥳
DoW operations and AI integration is one of the least discussed but highest-stakes enterprise deployments happening right now. The standard procurement cycle is completely mismatched with the pace of capability change. How are you seeing acquisition frameworks adapt to evaluate AI systems that could be functionally different by the time the contract is signed?
Honestly, I think most of the debate is anchored on the wrong question. Everyone’s arguing which LLM to license when the answer is which garden to plant in.
AI Foundry, Bedrock, Vertex, and the orchestration layers being built on top of them are what actually solve the procurement-vs-capability mismatch. You stop procuring a model and start procuring optionality, governance, observability, guardrails, and the ability to swap the backbone when a better one arrives. Contracts specify outcomes and SLAs, not model names.
The model becomes a substitutable component. The durable value sits in the layer that composes and governs it: orchestration, not decoration.
Really needed this. The constant AI hype is exhausting, and I agree completely! So much of the “design is cooked” and “you can make everything in minutes” discourse creates a false illusion of how designers should work, when taste, judgment, and real design thinking still matter most.
Another confirmation of Jevons Paradox.
If AI makes designers faster, we won’t need fewer designers. We’ll just produce more design.
The scarce part won’t be screens.
It’ll be judgment.
This! Should be written on a poster in any design studio
Finally a properly thought out take on this subject!
Thanks for that more nuanced look at your perspective on Claude Design. What I find interesting is that you still see Figma as part of the design production process. But isn't Figma exactly the tool where interaction states, edge cases, and viewports are the most cumbersome to represent, and where inconsistencies between design and code are most likely to happen?
Why shouldn't I develop my design system directly in a live repository at the code level in the future, instead of turning static images into code and vice versa? Personally, as a designer, I don't see much point in continuing to work in static layout environments when I'm designing digital and interactive products... For quick exploration or iterating through a few module variants—it's great for that. But for design system management, I don't see a future for the tool here.
Second point: If everything looks the same, how do brands and products differentiate themselves? I think the human comparison works well here: If everyone were equally good-looking, who would be more successful in the end? The person with the most pleasant personality and story. And I believe that applies to products and brands too. Design excellence is, as you correctly wrote, now free. Everyone can look good. And the point will come where AI is able to produce results that are less sloppy and more individual. But everything will predictably be pretty polished. I believe the only differentiator for brands can be their personality and story, rather than their visual appearance—because even that (in terms of brand design) is becoming increasingly similar and interchangeable.
Great article! I just did a smilar article https://mariamzai.substack.com/p/i-used-claude-design-and-this-is?r=88i190&utm_campaign=post&utm_medium=web
One more thing: when working alone whats your strategy for reiterating designs based on feedback? What kind of measurement tool and do you automate reports for it with some tresholds that will trigger a deeper investigation into a friction point ?
Interesting view, as a former art director/creative from the ad industry and a designer I see that these two branches is merging. What are your take on visuals, third party and native tools for creating content for webdesign? The combination of a designers intuition and the creatives ability to create different concepts in form of visualizations in form of animation, images and video will matter more than the layout. But that’s my two cents on the matter. Im not a seasoned designer so I’ve never followed a particular design system or anything like that.