AI is learning how your team works
Figma connects to your files, Canva edits existing designs, and Framer puts AI to work on A/B testing.
I've been noticing something interesting about AI tools lately.
They're starting to understand more of the work that happens around a design.
Not just the finished layout, but the research behind it, the brand guidelines it needs to follow, and what happens after something goes live.
That opens up some interesting possibilities, especially for teams that spend a lot of time moving information between tools or repeating the same production tasks.
Here's what caught my attention this week.
Figma can now pull information directly from Google Drive
This one feels pretty useful.
Figma added Google Drive as a connector for its AI agent and Make. That means you can bring research documents, spreadsheets, and images into a prompt without manually copying everything over.
You can also create a new Drive document from inside Figma, which could be useful for capturing design decisions or summarizing a review.
The bigger news is that connectors are now available on every Figma plan, including Starter.
I like the idea of giving an AI agent the actual research or content behind a design instead of asking it to work from a vague description.
There's one thing I'd check before connecting a client Drive, though.
Figma's documentation doesn't fully explain how its data-handling policies apply to content pulled through connectors. I'd review the permissions and training settings before giving an agent access to sensitive files.
Canva Brush can edit an existing design instead of starting over
Remember Canva Craft from last week?
Canva has introduced another in-house AI model called Brush, and this one focuses on editing designs you already have.
You can ask it to change a single line of copy, update elements across multiple pages, or reformat a design for another placement.
The interesting part is that the result stays editable in Canva.
If you've ever had to update the same headline across an entire presentation or resize a campaign for several channels, you can probably see where this could save time.
Canva says Brush is faster and more efficient than its previous design agent, although those numbers come from its own testing.
It's available to Canva Pro subscribers now.
I'd start by giving it a small, repetitive task and checking whether it leaves everything else alone. That's where these tools tend to become either genuinely helpful or surprisingly frustrating.
Framer's AI agent can now help run A/B tests
Here's an update I think web designers will appreciate.
Framer's Agent can now work with Convert, its A/B testing and analytics add-on.
You can ask the agent to create a test, set up a funnel, adjust targeting, or explain how an experiment is performing.
That could make it much easier to test something like a different hero headline or call-to-action treatment without spending so much time setting everything up.
I find the analytics side particularly interesting.
It's easy to launch an experiment and forget to come back to the results. Being able to ask an assistant what happened could make testing a more regular part of the design process.
Of course, I'd still check the actual data before making recommendations.
And this isn't a free feature. Convert is a paid add-on for Framer Pro and Enterprise sites.
Runway wants to turn your repeat workflows into small apps
This one is a little different.
Runway introduced App Studio, which lets you describe a tool in plain language and have Runway build and host it.
You could create something that generates product images using approved references, adapts creative for different formats, or handles another repeatable production task.
The interesting part is that you can connect these apps to your existing Runway workflows, assets, and Brand Kits.
Instead of asking everyone on a team to remember the right prompt and settings, you could give them a simple interface that handles the process consistently.
I can see this being useful for creative teams that need to produce lots of variations without reinventing the workflow every time.
There are some limitations worth checking first, though.
App Studio uses credits for building and running apps, and Runway's documentation isn't entirely consistent about sharing apps through public links.
I'd verify the costs and sharing options before building something a client or team needs to depend on.
Worth trying: make Firefly Boards part of your concepting process
I've been interested in how Adobe Firefly Boards can fit into actual creative work.
It's easy to use a tool like this as a place to collect references and generate images. But that can quickly turn into a giant board full of ideas that don't really go anywhere.
This week's workflow looks at a more structured approach.
Start with the creative brief, separate references by purpose, and use those references to guide your generations.
For example, you might have one group for composition, another for color, and another for materials or illustration style.
Then keep the strongest directions together so you're presenting a few thoughtful options instead of dozens of variations.
I especially like the idea of using the board to guide a conversation with a client rather than treating it as a gallery of AI output.
If you use Firefly Boards, this is worth trying on your next concepting project.
The Pulse: AI is starting to package design decisions into repeatable workflows
There's a pattern showing up across several tools this week.
Krea can look at a store's website, build a starting Brand Kit, and use it to generate a batch of ads.
Canva Brush can make changes across an existing design.
Runway can turn a creative workflow into an app that other people can use.
They're approaching different problems, but they have something in common.
The goal isn't just to generate a result. It's to make the process repeatable.
That's interesting for designers because so much of our work involves decisions that aren't visible in the final file.
Why a certain typeface works. Which colors belong together. How a product should be photographed. What makes something feel consistent with a brand.
These tools are starting to capture parts of that knowledge and apply it across multiple outputs.
I don't think that makes the designer less important. But it could change where we spend our time.
Instead of making every variation ourselves, we may spend more time defining the system, reviewing the results, and deciding where automation makes sense.
One more thing: Apple is opening Keynote to AI agents
Apple announced a set of updates coming to Creator Studio in November.
Pixelmator Pro is getting a generative fill feature, Final Cut Pro is incorporating MotionVFX tools, and Keynote, Pages, and Numbers are gaining support for outside AI agents through MCP.
The Keynote update is what caught my attention.
An AI assistant that can work directly with presentation files could be useful for updating decks, reorganizing content, or making repetitive changes across slides.
But there's an important distinction between an assistant that can edit a presentation and one that understands good presentation design.
I'd still want to review the hierarchy, pacing, and visual consistency.
These features haven't shipped yet, and Apple hasn't answered every question about data handling or availability.
Still, it's another sign that AI is moving closer to the files we're already working in.
That's it for this week.
I'll keep watching what's changing, testing what looks useful, and filtering out the stuff that probably isn't worth your time.
Until next Wednesday.