Shipping your code generation experiments to production
This video is featured in the Designing with AI 2025 playlist.
Summary
After this talk you may learn that:
- Trying out new AI tools, even if it means moving away from traditional ones, can help you build new skills and expand your capabilities.
- AI tools can speed up the early stages of design by quickly generating ideas, color palettes, and layouts, often faster than traditional design software.
- Working closer to actual code, with help from AI, allows you to create custom tools that support design decisions within the real context of your product.
Key Insights
-
•
Starting UI design from scratch is both an opportunity to discard old patterns and a challenge that requires rethinking milestones, team skills, and deliverables.
-
•
AI tools like ChatGPT and Claude can rapidly generate design elements such as color palettes and mock dashboards, significantly speeding up iteration compared to manual methods.
-
•
Utilizing AI to generate code-based prototypes allows designers to test UI concepts closer to real user experiences, reducing the gap between design and production.
-
•
The transition from designing in tools like Figma to working directly with code is facilitated by AI-assisted code generation, closing feedback loops and increasing designer agency.
-
•
Designers do not need deep coding expertise to begin integrating AI-assisted code generation; starting with small experiments, like text or color changes, is effective.
-
•
Collaboration and understanding the team’s diverse skill sets beyond job titles enrich the product development process and tool utilization.
-
•
AI-generated experiments should be presented as disposable starting points to encourage iteration and avoid false expectations about completeness.
-
•
Titles blending design and engineering, such as 'design engineer' or 'UX engineer', reflect the evolving nature of these roles in modern product teams.
-
•
Flexibility and openness to evolving AI tools is critical, as tooling rapidly changes and improves, requiring continuous learning and adaptation.
-
•
Deploying AI-generated designs and prototypes to actual production environments, as John did with Dino Deploy, accelerates real-world validation and team collaboration.
Notable Quotes
"I have somehow managed to not open Figma for the last six months, which I couldn’t have believed a year ago."
"Starting from scratch is really hard, but also a cool opportunity to throw away painful patterns."
"One of the most important questions before starting is what tools do I have to work with."
"I thought, maybe now is the time for change, maybe designers should code—maybe the better question in 2025 is should I code."
"With AI tools, you have to come back and reprompt, it’s not a one-shot operation."
"I generated a Tailwind color palette with AI that took seconds, compared to what used to take me some time manually."
"Having tight feedback loops and directly controlling the end-to-end experience completely changed how excited I was to do design work."
"If you never want to fully embed yourself in engineering, that’s fine too — you don’t have to show anyone the code you generate."
"Throwing away old methods for something new can unlock skills you didn’t know you had."
"AI tools have never made it easier to add flourishes like animations or excitement to your UI."
Or choose a question:
More Videos
"Begin every research project with the end in mind by framing what impact you hope to drive."
Roberta Dombrowski Sam Duong WoloszynskiMaking Research a Team Sport
March 11, 2022
"Nobody loves it when someone from another department shows up to tell people how to do their jobs better."
Laura KleinUnique challenges of innovation in enterprises
April 23, 2020
"Sometimes you have to give permission to yourself and your team to try something different."
JD Buckley Margot Dear Jim Kalbach Janaki KumarCOMMUNICATE: Discussion
June 14, 2018
"Investors front load the cost they expect to win at all costs."
Jen van der MeerService design performs value
November 19, 2025
"Most bad panels feel like a bunch of people showing up at the last minute without shared goals or preparation."
Louis Rosenfeld Jemma Ahmed Christian Crumlish Uday Gajendar Chris GeisonCoffee with Lou #3: What Makes for a Successful UX Conference Presentation?
May 2, 2024
"What if design ops’ job was to condition the environment and remove friction?"
Jacqui FreyFlow and Superfluidity for Design Orgs
November 7, 2018
"People form habits on software that become efficient, so they naturally resist changes that force new habits."
Paula BachImproving Legacy Software: How Much Better Does it Have to Be?
March 11, 2022
"Alert fatigue is real; users can't be burdened with constant security decisions or they'll ignore them."
Heidi TrostWhen AI Becomes the User’s Point Person—and Point of Failure
August 7, 2025
"We’re not doing this for the good of research. Research is merely a means to an end."
Andy WarrUnder My (Research) Umbrella: The Benefits and Challenges of Building a Unified Insights Function
March 25, 2024
