Imagine you design a brand, add a single product, press one button, and a few minutes later you have product photos, social banners, website visuals and even a looping hero video. No juggling five AI tabs, no dragging files back and forth. That is exactly what a well-built Figma Weave workflow can do, and in this article I’ll walk you through how I set one up for a fictional matcha shop.

By the end you’ll know how to chain AI models together, connect them to your real Figma designs, and use prompt variables to get dozens of variations out of one workflow. For me, the whole system saved about 5 to 7 hours of work.

If you’ve never opened Weave before, start with my Figma Weave tutorial, where I cover the basics and add an AI video to a website hero.

Why a Figma Weave workflow beats generating images one by one

Most of us use AI image tools like slot machines: write a prompt, get an image, download it, drop it into a design, notice the background is wrong, repeat. That’s fine for one asset. It falls apart the moment you need a whole campaign.

Figma Weave is a node-based AI platform, so instead of generating things one at a time you build a visual workflow. A prompt node holds your text, a model node generates the image, a tool node removes the background. Nodes on the left are inputs, nodes on the right are outputs, and everything flows from left to right. Once the chain exists, you can run all of it with a single click.

The other big win is variety of models. Weave gives you image models like Nano Banana, video models like Kling and Seedance, plus 3D and audio tools, all in one place. You don’t need a separate account for each one.

Step 1: Start with a real brand setup in Figma

A workflow is only as good as what it plugs into. Before touching Weave, I prepared a small design system for the matcha shop: a simple logo, vertical social banners, story banners, a menu and a website. Every layout was finished except for one thing, the images. That’s the point. Weave’s job is to fill those empty slots with on-brand content.

One tip: if you want to feed your logo into Weave, use a regular frame rather than a component. I had to detach mine first.

Step 2: Connect your Figma and Figma Weave accounts

Figma and Figma Weave are two separate accounts, so the first move is to open Weave’s settings and link them. Without that link you can’t sync assets with your Figma files.

Once connected, create a new file in Weave. This is your canvas for the whole project.

Step 3: Build the core Figma Weave workflow

Here’s the simple chain behind every product image:

  1. Prompt node. Write the prompt for your first drink. I used four prompts, one per matcha cup.

  2. Image model node. Drag from the prompt (or right-click and search) to add Nano Banana, which generates the image from your text. I compared it with other image tools in my Nano Banana comparison.

  3. Remove background node. Search for “remove background” and connect it so each cup comes out cut out and ready to place on any banner.

Repeat for each drink and you have a small production line for product shots.

Step 4: Sync your Figma banners with Weave

This is the part that surprised me most. Copy your banners from Figma and paste them straight into Weave. Zoom in on a card and you’ll see a Connect button. Click it, pick your Figma file, and Weave pastes the banner into your work file as a regular frame, marked with a small Weave icon next to its name.

That icon means the frame is synced. If you change something in Figma, like a color or a text alignment, the button in Weave turns into Update, and one click pulls those changes in. Do this for every banner you want to automate.

Next, link each image slot on a banner to the generated image that should fill it. Click the banner card, use the connect icon in the input panel, and choose the image node. Hover over a slot and you’ll see a blue line showing which generation feeds it. A banner with two images simply gets two links.

Step 5: Generate brand photography with your logo and a prompt enhancer

Product cutouts are only half the story. For lifestyle imagery, I added more Nano Banana cards with their own prompts, such as a young woman holding a matcha cup. I linked the logo and the product images as references so the model uses the exact cup design.

For trickier ideas, let AI help you prompt. I wanted a floating matcha drink with strawberries on a plain pink background but wasn’t sure how to describe it. So I dictated a rough sentence, passed it through a prompt enhancer node to get a richer prompt, and fed that into the image model along with my reference images. Labeling them as image one and image two in the prompt keeps the references clear.

Step 6: Run the whole workflow with one click

Select all your nodes and Weave shows which models will run and the total credits they’ll burn. Each generator card also shows its own cost. For example, Nano Banana at 1K resolution costs 7 credits, while 4K costs 13. For website visuals, 1K was enough for me.

Then hit Run. The product images generate, backgrounds disappear, and the results land directly inside your connected banners. Because those banners are synced, the finished images are already waiting in your Figma file. From there it’s regular design work: change a font color to white, add an overlay, crop the composition, scale things up.

Step 7: Use prompt variables to create dozens of variations

Variables are my favorite part of a Figma Weave workflow. In a prompt, you can turn any phrase into a variable and attach a text iterator that holds a list of values.

My example was a photo of a young couple on a rooftop, each holding a different matcha. I made one variable for the first person’s drink (strawberry and banana) and one for the second person’s (ube and classic). Weave then offers two ways to run it:

  • Parallel pairs the values one to one. In my case that meant two images for 14 credits.

  • Matrix combines every value with every other value. That meant four images for 28 credits.

I used the same idea for a billboard mockup. One prompt placed my real banner on a vertical billboard in Soho, New York, with a time-of-day variable: sunny summer day, golden hour, fall evening and snowy winter. One click, four scenes, all using the actual design from Figma. Perfect for a portfolio or a brand book.

Step 8: Turn your images into video

A static hero is nice. A moving one is better. I took my favorite billboard shot and first ran it through a color correction node to make it warmer and more saturated. Color correction is free, since you only spend credits when something is generated. I also re-ran it with the logo added to the cups.

Then I tested two video models. Seedance cost 281 credits, while Kling cost only 41, so I went with Kling for most runs. A lesson learned: my first video, made with a prompt enhancer, went strange and even broke the cup. When I removed the enhancer and used my own simple prompt, the 10-second result looked great and even came with sound. Sometimes a short, clear prompt beats a clever one.

Back in Figma, I dropped the video into the website hero as a background, muted it, set it to loop and added a dark fill so the text stayed readable. In prototype mode the page feels alive. If you want to build the website itself with AI, here’s how I built an award-winning website with Claude.

Why this Figma Weave workflow matters for designers

The real value is what the whole system does for you:

  • It automates the repetitive work. Generating images, removing backgrounds, placing them into banners and syncing with Figma all happen in one run, with no downloading and re-uploading between tools.

  • It becomes a content and brand system. Your logo, banners, prompts and reference images live in one workflow, so every new asset follows the same look. Add a new drink, run the sequence again, and it flows into all your templates.

  • It saves serious time on content production. In my case, it replaced about 5 to 7 hours of manual work, and you can share the workflow with your team or publish it to the Figma community.

Tips to get more out of your Figma Weave workflow

  • Design first, generate second. Build your banners, menu and website in Figma before running anything, so every image has a place to land.

  • Check credits before you run. Select your nodes and read the total. Dropping Nano Banana from 4K to 1K, or choosing Kling over Seedance, makes a big difference.

  • Choose parallel unless you need every combination. Matrix mode doubled my cost in the rooftop example.

  • Keep video prompts simple. My plain prompt beat the prompt enhancer for video.

  • Re-sync often. Press Update in Weave after tweaking layouts in Figma, and change the order of image links if two banners need matching compositions.

Conclusion: your Figma Weave workflow is a content system, not a one-off

What started as a single product image became a full content production system: cut-out product shots, brand photography, social banners, a billboard mockup in four seasons, and a looping hero video, all connected to real Figma files. That’s the real power of a Figma Weave workflow. You stop generating assets one at a time and start building a machine that makes them for you.

Want to see every click? Watch the full tutorial on YouTube and subscribe to my newsletter so you don’t miss the next AI design workflow. You can also grab my free resources or start learning on my learning page.

If you’d like to build your own workflow, you can try Figma Weave here.