Claude Design is slow and I love it anyway (plus why I love ChatGPT Images 2.0)
Claire Valle reviews Claude Design's capabilities for importing design systems and creating marketing prototypes, while also testing ChatGPT's new Images 2.0 model for brand kit creation and layout work. She finds both tools promising but notes significant speed limitations and credit restrictions that impact iteration workflows.
Summary
Claire Valle provides a comprehensive review of two major AI design tool releases. She extensively tests Claude Design by importing design systems (using Lenny's Newsletter as an example) and finds the 5-minute design system extraction process impressive, though slow. She demonstrates three main use cases: creating marketing landing pages that adhere to brand guidelines, generating slide decks from existing content, and creating playful redesigns without design system constraints. Valle particularly praises Claude Design's ability to maintain brand consistency and generate high-quality copy, but criticizes the slow feedback loops compared to traditional tools like Figma. She also explores ChatGPT's new Images 2.0 model, testing its ability to create multi-page brand kits and perform personal color analysis. She notes significant improvements in typography and layout capabilities, though the model sometimes produces facial distortions. Valle mentions Google's new design.md standard for describing design systems to AI tools. Throughout both reviews, she emphasizes the frustration with usage limits and credit systems that interrupt workflows, while acknowledging the tools' potential for marketing teams and rapid prototyping.
About this episode
In this mini episode, I do a full walkthrough of the AI design tools that dropped in April 2026: Anthropic’s new Claude Design, OpenAI’s GPT Images 2.0, and Google Labs’ open-source DESIGN.md format. I import a full design system from Lenny’s Newsletter, build a landing page, turn my own article into a polished deck, generate a brand kit for ChatPRD, and run a personal color analysis from a photo. *What you’ll learn:* 1. How Claude Design handles design system imports and whether it can actually replace Figma 2. The three best use cases for Claude Design: marketing landing pages, slide decks, and creative redesigns 3. Why ChatGPT Images 2.0 is a breakthrough for brand kits and layout work 4. Google’s new DESIGN.md standard 5. The practical limits of AI design tools (spoiler: you’ll hit credit limits fast) *Brought to you by:* WorkOS—Make your app enterprise-ready today: https://workos.com?utm_source=lennys_howiai&utm_medium=podcast&utm_campaign=q22025 Rippling—Stop wasting time on admin tasks, build your startup faster: https://rippling.com/howiai *In this episode, we cover:* (00:00) Welcome and what’s in the spring 2026 AI design drop (01:45) Claude Design overview (03:05) Importing Lenny’s Newsletter design system into Claude Design (04:06) How Claude Design structures a design system (05:42) Google Labs’ DESIGN.md standard (06:41) Building Lenny Doc, a PRD generator landing page using the Lenny design system (09:44) Why the three-variation output is Claude Design’s smartest UX choice (10:20) Hitting the Claude Design limit and paying $200 to keep going (11:05) Where Figma still wins (13:20) Reviewing Lenny Doc (16:19) Turning an Open Claude article into a branded slide deck (17:57) The ’90s GeoCities “Lenny’s Product Zone” redesign (19:44) Claude Design recap (20:15) ChatGPT Images 2.0 and what makes it the first “thinking” image model (21:25) Generating a multi-page brand kit for ChatPRD and iterating with reference images (23:43) Personal color analysis demo (26:02) Recap *Tools referenced:* • Claude Design: https://claude.ai/design • ChatGPT Images 2.0: https://openai.com/index/introducing-chatgpt-images-2-0/ • Midjourney: https://www.midjourney.com/ *Other references:* • Google’s DESIGN.md: https://stitch.withgoogle.com/docs/design-md/overview • Lenny’s Newsletter: https://www.lennysnewsletter.com/ • Jamie Gannon “How I AI” episode on reference styles: https://www.lennysnewsletter.com/p/mastering-midjourney-how-to-create • Brand prompt inspiration: https://x.com/riomadeit/status/2046682442791071787 • Figma team “How I AI” episode on design systems: https://www.lennysnewsletter.com/p/from-figma-to-claude-code-and-back *Where to find Claire Vo:* ChatPRD: https://www.chatprd.ai/ Website: https://clairevo.com/ LinkedIn: https://www.linkedin.com/in/clairevo/ X: https://x.com/clairevo _Production and marketing by https://penname.co/._ _For inquiries about sponsoring the podcast, email [email protected]._
Key Insights
- Valle argues that Claude Design makes design systems a first-class citizen, addressing the number one complaint about prototyping tools never matching existing brand guidelines
- Valle claims Figma will continue to win because design tools require really short feedback loops, and the slow LLM cycles in Claude Design create iteration bottlenecks
- Valle observes that Claude Design performs better at marketing-style landing page components than UX design and app components when working with design systems
- Valle identifies that Claude Design writes exceptional copy as an underappreciated strength, calling it a secret power of the tool for building great prototypes
- Valle notes that ChatGPT Images 2.0 is the first image model that can do thinking and has significantly improved typography without the classic AI text tells
Topics
Transcript
[0:01] [music] >> Welcome back to How I AI. I'm Claire Valle, product leader and AI obsessive, here on a mission to help you build better with these new tools. This [music] week, there has been so much released in the AI for design space. Whether you're looking at Claude design, trying to decide if it's really going to replace Figma, or want to know if the new GPT image 2 model is actually that good, I've got you covered with a mini app where I'm going to show you what I really [music] think about these new design tools and how they'll work for [0:31] your business and a couple fun use cases. Let's get [music] to it. This…
Full transcript available for MurmurCast members
Sign Up to AccessMore from How I AI
GPT-5.6's video editing via Codex is genuinely one of my favorite new workflows
A product manager describes using GPT-5.6 with Codex to automate video editing for social media clips. By simply dragging a file and providing natural language instructions, the AI generated five polished, fast-paced hype videos from a long conference talk recording, dramatically reducing the time-intensive manual clipping process.
Theoretically Intelligent vs. Practically Effective: Why GPT-5.6 Sol Beats Fable for Product Work
An executive contrasts two AI models (Fable and Soul/GPT-5.6 Sol), arguing that Soul is superior for product work because it prioritizes practical effectiveness over theoretical intelligence. The speaker values the ability to ship products to customers and understand end-user goals over theoretical sophistication.
Build a harness when the same workflow needs the same setup and the same outcomes, every time
Building a harness for repetitive workflows allows you to be more prescriptive about job execution, resulting in greater efficiency, consistency, and better outcomes. Rather than explaining requirements to an AI agent each time, a harness lets you use a simpler interface like pasting a link while the agent already understands the intended task.
GPT 5.6-Sol vs. Claude Fable: Why OpenAI’s new model crushes my benchmark
Claire Vo compares OpenAI's new GPT 5.6 models (Soul, Terra, Luna) against Claude's Fable using her custom "How I AI" benchmark, finding that GPT 5.6 Soul excels at practical product work, prototyping, and natural communication, while Fable is theoretically intelligent but pedantic and difficult to collaborate with.
Context offloading is an underrated AI use case
The speaker highlights context offloading as an underrated AI use case, where AI serves as a safety net for routine cognitive tasks like email management and personal finances. Rather than adding new capabilities, AI reduces anxiety about missing important information or making mistakes by handling monitoring tasks, thereby freeing up mental bandwidth.