Automated Digital Design Frameworks

Traditional surrogate-based design optimization (SBDO) is hitting a wall, especially with high-dimensional, complex designs. In this new paper, Dr. Namwoo Kang presents a next-gen framework using generative AI, integrating three key models: - Generative model (design synthesis) - Predictive model (performance estimation) - Optimization model (iterative or generative) Rather than optimizing directly in a high-dimensional design space (x), the workflow introduces a low-dimensional latent space (z) learned via generative models. ➡️ z → x → y z = latent variables x = CAD geometry y = performance (drag, stress, etc.) This means we’re no longer hand-coding design parameters or doing trial-and-error with simplified surrogate models. 🧠 Why this matters: - Parametric modeling is no longer a bottleneck - Complex shapes are learned directly from CAD - Dynamic and multimodal performance data (1D, 2D, 3D) can be used - Near real-time optimization is possible #AI #GenerativeDesign #CAE #DesignOptimization

Your design system documentation has a 3-week lag problem 👇 Designer updates the button → Developer ships it → Someone hopefully remembers to update the docs. The result? 🤯 → "Is this the latest version?" 12 times per sprint → Hours wasted hunting for correct specs → 30% of components still using old tokens months later Most teams try to solve this with better processes. More meetings. Stricter update cadences. Automated reminders. That's optimizing the wrong thing. The only way to kill latency is to connect your tools so they document themselves. ✨ Here is the automated design system documentation workflow: Figma (API + MCP) → AI reads specs (I used Claude Code) → Mintlify auto-deploys What gets automated: → Screenshot exports from Figma frames → Spec extraction (spacing, colors, tokens) → Documentation updates → Pull requests with visual diffs ✨ You can even set up GitHub Actions to check tracked Figma frames weekly and create PRs automatically. The guide is available on today's newsletter. 🙌 What's your setup? #designsystem #documentation #productmanagement #productdesign

The next evolution of design systems isn’t visual. It’s behavioral. We’re moving beyond static component libraries and theme tokens — toward dynamic, intent-aware experience frameworks that adapt to context. Recent Figma updates are accelerating this shift: Variables + Modes — Tokens are no longer just colors or spacing. They’re becoming logic — expressing states like risk=high, density=compact, or mode=investigate. Dev Mode + Code Connect — Design and code now share a single truth, collapsing the gap between “spec” and “ship.” MCP access for AI agents — Agents can now read structured design data, understand context, and even apply changes directly. This opens the door to living systems: - Components that shift tone or density based on risk or urgency. - Layouts that adapt automatically to task context. - Policies that guide what an AI assistant can safely modify. In short: design systems are becoming autonomous frameworks for human-machine collaboration.

Most of the AI-meets-design conversation right now is about converting. Designs to code. Code to designs. Back and forth. It's great. But what about creating? I started with 4 things: → A blank Figma canvas → Claude Code → Figma Console MCP → Material 3's component library One prompt: "build a mobile fintech login screen using the existing components and tokens." Claude analyzed the full design system, picked the right components, set the right properties, and composed the layout directly on the canvas. Real components, real variables, fully bound to tokens. But I didn't stop there! THEN, I asked it to invent a Brutalist theme. → It spun up one of our custom UI designer sub-agents → Created a new variable mode from scratch (acid yellow, zero radii, Space Mono) → cloned the original layout, and restyled everything Same components, completely different look/feel. Switch modes and it all holds together. 15 minutes. Start to finish. The magic is how to stack tooling, not a single tool. MCP for the canvas, Claude Code for orchestration, sub-agents for specialized design thinking, and a solid design system underneath it all (very important). This is a creative tool, not just a conversion tool. Style exploration, mood boards, rapid variable mode testing, pushing your token architecture to see what it can handle... I did this in 15 minutes. I want to see what you can do in an hour. Grab the Figma Console MCP, plug in your design system, and show me! If you need help getting set up or want to talk about making your design system AI-ready, reach out. Check out the new easy-to-follow community setup guides - https://lnkd.in/eNmzhh5S

The market pressure is real. By 2027, about $130 trillion is expected to flow into capital projects, even as productivity has only ticked up around 1% compared to 3.6% in manufacturing. Typical overruns reach roughly $1.2 billion with delays from six months to two years, and margins hover near 5%. Research shows top performers lean into platform, modular, and rules-based design. They’re more likely to automate quotes and use design automation, which helps them move faster while controlling risk.     If your teams are stuck translating bespoke requirements through siloed tools and manual steps, you feel the strain fast. Long lead times, margin-eroding errors, and penalties for late delivery stack up. I’ve seen the same pattern across capital assets. When engineering is the bottleneck, quoting and ordering slow to a crawl.     There’s a way to change the shape of the work. Industrialize the design process. Build modular platforms that are standardized yet configurable. Then layer rules-driven design automation on top. Capture the design rules once, reuse them across orders, and automatically generate the outputs your downstream teams need. Think BOMs, 3D models, and drawings produced with the same speed and precision you expect from standardized products. That shift reduces unique upfront engineering, protects quality, and frees specialists to focus on the hard problems.     Want to cut through complexity? Do this, pick one asset family. Map the core design rules that drive 80% of variation. Connect those rules to CAD so the system auto-generates BOMs and drawings for your two most common configurations. Run it for 30 days and track cycle time, rework, and the number of manual handoffs removed. If the signal is positive, expand.     If this is your world, what’s the first rule you’d automate to remove a bottleneck? 

New research out of Department of Mechanical and Process Engineering (D-MAVT), ETH Zurich on AI Agents and CAD, offering a framework for generative geometry. "From text to design: a framework to leverage LLM agents for automated CAD generation" by Aurel Schüpbach, Raúl San Miguel Peñas, Julian Ferchow and Mirko Meboldt introduces a CAD and LLM agnostic framework to evaluate different agents and compare their performance. For their examples, the agent with vision capabilities was the most sucessful. What I find most interesting is the automated topology optimization, run with Grasshopper of course, and the tOpos plugin. Cutting edge capabilities live in Rhino! link to the paper below!

New spatial representations are needed to be able to integrate automated design and AI-driven design workflows for architecture and the built environment. In my research, I created a new graph-based method for representing floor plans as a hypergraph. With this new framework we can represent spatial connectivity, area and geometric configuration in a single graph-based data structure. The hypergraph can then be used to automatically analyze, represent and generate new architectural floor plans... connecting this with building performance simulations opens up a new world for better understanding and creating low-carbon buildings! Learn more in the link to the blog post that summarizes the work I just published together with #MIT PhD co-advisors Christoph Reinhart and Caitlin Mueller on the Nature Blog: https://lnkd.in/er8GNVnu #architecture #buildingperformance #floorplan #ai #generativedesign #design #algorithm #MIT #Berkeley