This is great, have to echo what Neural Foundry said about the 'divergent vs convergent' comment, which is so so true. It is also something we as designers can easily get behind - double diamond anyone?
Hey Sherizan — the divergent vs convergent framing is clean and practical. "Don't polish before exploring" is the kind of rule that sounds obvious but I watch people violate constantly — they get one AI-generated mockup and immediately start tweaking pixels instead of generating ten more options first.
I build AI agents for businesses and the same two-mode thinking applies to agent design. There's a divergent phase —what could this agent do? What personality should it have? What workflows could it handle? And then a convergent phase — okay, now let's scope it tight, define boundaries, lock the voice. Teams that skip divergent mode build something too narrow. Teams that never converge ship something that tries to do everything and does nothing well.
The prompt-writing-as-core-skill point is underrated too. The people getting the best results from AI tools aren't the most technical — they're the ones who can articulate what they want clearly. That's a design skill whether you're prompting Figma or scoping an agent. Good map.
Could you please explain „Step 4: Iterate between Figma ↔ Cursor
This loop has become the real magic.“ In more detail? I know there is a lot of content around that, but I am still confused. How would I export from Figma to cursor?
Great points! Not really sure about the statement "Stitch for ideation/concepts with no friction", because my experience with Stitch creates more friction each time I give it a try.
This is great, have to echo what Neural Foundry said about the 'divergent vs convergent' comment, which is so so true. It is also something we as designers can easily get behind - double diamond anyone?
I look forward to more great insight!
Hey Sherizan — the divergent vs convergent framing is clean and practical. "Don't polish before exploring" is the kind of rule that sounds obvious but I watch people violate constantly — they get one AI-generated mockup and immediately start tweaking pixels instead of generating ten more options first.
I build AI agents for businesses and the same two-mode thinking applies to agent design. There's a divergent phase —what could this agent do? What personality should it have? What workflows could it handle? And then a convergent phase — okay, now let's scope it tight, define boundaries, lock the voice. Teams that skip divergent mode build something too narrow. Teams that never converge ship something that tries to do everything and does nothing well.
The prompt-writing-as-core-skill point is underrated too. The people getting the best results from AI tools aren't the most technical — they're the ones who can articulate what they want clearly. That's a design skill whether you're prompting Figma or scoping an agent. Good map.
Thanks!
Love it!
Thank you Sarah
I did a bunch of work with Cursor in the past few days and it feels like I want to stop to rest. These productivity spikes are killing me :)
Love these. Especially Design Systems matter more now than ever. In large orgs that don’t have established systems is going to be like herding cats 🐱
This is really insightful. Thanks
Could you please explain „Step 4: Iterate between Figma ↔ Cursor
This loop has become the real magic.“ In more detail? I know there is a lot of content around that, but I am still confused. How would I export from Figma to cursor?
Great points! Not really sure about the statement "Stitch for ideation/concepts with no friction", because my experience with Stitch creates more friction each time I give it a try.
You only need Claude Code. You don’t need anything else. Maybe Agentnation.
So one prompt for each px size change? Idk if that is the most efficient way to design something high precision….
What tool are you still missing as a designer?