The Brief: AI is getting easier to control
October 7, 2026
I noticed a theme in a lot of the AI updates this week.
We’re getting more ways to actually control what AI makes.
That sounds obvious, but it’s been one of the biggest problems with these tools. Getting a good first result is one thing. Keeping a style consistent, making a small edit without breaking everything else, or getting something into a real production workflow is much harder.
A few updates this week seem to be working on exactly that.
Here’s what I think is worth knowing.
Figma Motion is starting to look a lot more useful for real projects
Figma Motion added custom animation styles, audio, text animation, and Lottie export.
The custom styles are what caught my attention. You can create reusable motion styles and publish them to a library, which starts to make motion feel more like part of a design system instead of something every designer recreates on their own.
Lottie export also makes this much more interesting for work that eventually needs to leave Figma and make it into a product or website.
Ideogram is trying to solve the "one more edit" problem
Ideogram 4.5 is built around iterative editing.
The goal is to let you change one part of an image repeatedly without everything else slowly changing along with it. If you’ve spent any time editing AI-generated imagery, you know how quickly that drift can make a promising image unusable.
I’m interested in this less as another image model and more as another sign that AI image tools are moving past generation and into actual editing workflows.
Meta's Muse could change who is doing everyday design work
Meta introduced Muse for Small Business, an AI agent that can connect to tools including Canva, Figma, and Shopify.
That means a small-business owner could potentially ask an agent to work across the tools they already use instead of opening each one and doing the work themselves.
For designers, I don't think the interesting question is whether this replaces design. It's what happens to the smaller production tasks that clients currently need help with.
If an agent can handle some of those jobs, our value moves even more toward the systems, judgment, direction, and original thinking behind the work.
Worth trying: Figma's shaders
This is one of those Figma features that’s easy to miss.
You can create custom shader fills and effects with a plain-language prompt, including animated and cursor-reactive effects.
The interesting part is that you don't need to know shader programming just to experiment. You can describe what you're after, then keep refining it from there.
I wouldn't use an effect just because AI made it easy to create one, but for interactive web work, this feels like a genuinely useful playground.
The Pulse: Adobe owns Topaz Labs now, and people are watching what happens next
Adobe completed its acquisition of Topaz Labs on September 23.
Topaz says its products will remain standalone. But across creative communities, people are understandably watching what happens with pricing, licensing, and credits.
There isn't a dramatic conclusion here yet, and that's probably the point.
Nothing needs to change immediately for an acquisition like this to matter. The useful thing right now is simply knowing it happened and keeping an eye on what Adobe does with Topaz from here.
One more thing: AI can extend a shot, but check the join
I took a closer look at Generative Extend in Premiere Pro this week.
It can be genuinely useful when a clip ends just a little too early. Think a few extra frames for a transition, enough time to hit a music cue, or some extra room tone.
The catch is that generated extensions can introduce brightness changes, flicker, softness, or other artifacts that aren't always obvious in the timeline.
This feels like a good example of where AI fits right now: give it a small, specific job, then check its work.
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.
See you next Wednesday.