8) Ask Claude to double-check its work, look for edge cases, suggest improvements, and ensure nothing broke.
9) Remember that Claude can research and browse the web for additional context or documentation.
6) Use the `/compact` command when you hit milestones to compress down your context. If you don't, it'll force you to do it anyway.
7) Ask it to break down large tasks and run subagents simultaneously for faster completion.
4) Since there's no restore function, push to your repo often at key checkpoints so you always have a backup.
5) If you're using VS Code or JetBrains, use the accompanying extension for better functionality.
3) Run `/init` to automatically create the CLAUDE markdown file. Ask Claude to "remember" something and it will add it to this file. Put all your rules (think Cursor rules) here.
1) Press Shift-Tab to get into "Plan" mode. Use that generously to figure out what you're going to have the AI do before it does it.
2) Drag reference images directly into the terminal.
I've been blown away by the quality of work that Claude Code produces compared to Cursor, and I want to give you 9 practical tips to get the most out of it:
As execution compresses, clarity becomes the real deliverable.
And that's where designers can shine.
Our real value goes beyond what's on the screen:
→ Framing things from the user's perspective
→ Asking the right questions
→ Bringing research into the room
→ Deep expertise in experience
→ Strategic thinking about what to build and why
The design file is still useful, but it's no longer *the thing*. It's more like a shared map, a conversation starter, and a snapshot of what we're trying to accomplish.
Designers are in a unique position to lead here. We can shape direction, create clarity, and help teams align around the right problem and the best solution.
It's about direction.
Where are we headed?
Why are we building this?
How do we make sure everyone's aligned?
That's becoming more important than almost anything else.
As more people get the tools to build, and as the building itself gets faster (thanks to AI and code-first workflows), the center of gravity is shifting.
It's not just about shipping polished Figma files anymore.
This release — and this Config overall — makes it feel like Figma's not just a design tool anymore. It's becoming something bigger, and I'm curious where it goes next.
Things I'm still hoping for:
→ Let me use AI directly in the editing canvas, so I can leverage it while designing and utilize my design system.
→ Open it up to community plugins — we'll come up with use cases they haven't even thought of yet.
4. Better dev handoff. Since Figma already knows everything about your design file, the AI-generated code has the opportunity to be way more accurate. And you're handing off a prototype, not just static screens.
3. AI that actually feels useful. You're not just prompting a black box — point to specific parts of your design and tell the AI what to do. Then iterate in a shared conversation with your team. Feels more like working with the AI instead of just hoping it gets things right.
2. Easier, better feedback loops. You can publish prototypes straight to the web (custom domains, permissions, analytics, etc.). It's fast enough that you don't have to think twice about getting feedback earlier.
How Figma Make is changing things for product designers:
1. It's all in one place now. You can go from design to working prototype without leaving Figma — and even pull in the designs you're already working on. No extra tools, no weird imports/exports.
From what I’m seeing in products like Subframe, Onlook, Tempo, and Paper, I’m convinced we’re getting close.
But we’re not there yet.
7) AI assistance throughout — offering micro design critiques, accessibility help, idea generation, nudges from the design system and past work, hints from user research (which it also helped synthesize), and more.
5) Easy publishing of the prototype to a unique URL so stakeholders can interact with it without breaking anything.
6) A seamless handoff: submit designs directly to a branch so engineers can review and create a merge request.
3) The ability to design complex screens and multi-state flows — using realistic dummy data copied or generated from my app — making precise edits in a tool just as powerful as Figma.
4) Ability to create a prototype that's built on my app’s real code from my designs.
🧵 What I want from an AI product design app (this doesn’t exist yet):
1) A place to build and utilize my official design system.
2) Direct connection to my product’s real codebase — with changes submitted via branches and merge requests.
Product design has already been shifting from process obsession to faster shipping, more iteration, and closer dev collaboration. Subframe seems like it could supercharge that.
We need more tools like this.
If you're a designer or PM, this is worth checking out.
www.subframe.com
You can make a design, push it most of the way to engineering-ready code, and hand it off within your actual codebase. 3/4
Unlike v0, Lovable, Bolt, and other tools that churn out a generic UI, Subframe digs into the task of working with engineering to edit and deliver designs without compromising UX thinking. 2/4
Having a real "my job will never be the same" moment.
I've been exploring Subframe and chatting with a designer about what he's testing — and this might be the first AI design tool I've seen that will actually disrupt the everyday work of product design. 1/4